Predefined space-time pitch angle control method of variable-speed wind generating set

Abstract

The invention discloses a predefined space-time pitch angle control method of a variable-speed wind generating set, and belongs to the technical field of variable-speed wind generating set control. The method is characterized by comprising the following steps of 1, obtaining a dynamic model of the variable-speed wind turbine according to the Betz theory; step 2, obtaining a predefined space-time reaching law with a buffer area according to the improved barrier function; 3, constructing a sliding mode variable, and converting a non-affine model of the variable speed wind turbine into an affine model based on an unknown smooth nonlinear function; and 4, compensating the unknown smooth nonlinear function by using a neural network, and obtaining a pitch angle controller according to the neural network. According to the predefined space-time pitch angle control method of the variable-speed wind generating set, the state space is divided into the multiple parts including the buffer areas, dependence on model parameters is reduced, the buffeting phenomenon is weakened through the improved obstacle function, and then the power generation efficiency is improved, and the operation cost is reduced.

Description

TECHNICAL FIELD

[0001] The application relates to a predefined spatiotemporal pitch angle control method of a variable-speed wind turbine generator set, and belongs to the technical field of variable-speed wind turbine generator set control. BACKGROUND

[0002] A wind turbine is a kind of turbine using wind energy as power, and a variable-speed wind turbine has become one of the most rapidly developed wind turbines in recent years. The variable-speed wind turbine adjusts the rotor speed in real time to adapt to wind speed changes, and significantly improves the wind energy capture efficiency. When in a full-load region, the control target is changed to maintain the output power at a rated value by adjusting the pitch angle.

[0003] For the control of the pitch angle of the variable-speed wind turbine, the following methods are mainly included in the prior art:

[0004] 1. Documents: Uehara, Akie and Pratap, Alok and Goya, Tomonori and Senjyu, Tomonobu and Yona, Atsushi and Urasaki, Naomitsu and Funabashi, Toshihisa. A coordinated control method to smooth wind power fluctuations of a PMSG-based WECS [J]. IEEE Transactions on energy conversion, 2011, 26 (02), 550-558 and document: Stol, KA and Fingersh, LJ. Wind turbine field testing of state-space control designs [J]. Golden, CO: National Renewable Energy Laboratory, NREL / SR-500-35061, 2004. The proportional integral (PI) and linear quadratic Gaussian strategies are used for the control of the pitch angle. However, the dynamic characteristics of the wind turbine in the full-load region are typical uncertain nonlinear systems, and the dynamic model has the following characteristics: firstly, the aerodynamic torque and the pitch angle present a complex non-affine relationship, and the traditional linear control method is difficult to effectively handle; secondly, the system parameters (such as air density and blade surface roughness) will drift with the change of environmental conditions; and thirdly, the rate limit and dead zone characteristics of the actuator further increase the complexity of the control design.

[0005] These factors collectively result in a strongly coupled, time-varying system dynamic response, thus imposing extremely high requirements on the robustness and adaptability of the control algorithm. In response to these challenges, current research has mainly focused on the development of advanced control strategies, including adaptive control, model predictive control (as in Petrovic, Vlaho and Jelavic, Mate and Baotic, Mato. MPC framework for constrained wind turbine individual pitch control [J]. Wind Energy, 2021, 24(01): 54-68.) and intelligent hybrid control (as in Jiao, Xuguo and Meng, Wenchao and Yang, Qinmin and Fu, Lingkun and Chen, Qi. Adaptive continuous neural pitch angle control for variable-speed wind turbines [J]. Asian Journal of Control, 2019, 21(04): 1966-1979.) and other methods to balance the control accuracy, robustness and real-time requirements.

[0006] However, most of the control methods proposed in the existing literature can only ensure the asymptotic stability or the uniform ultimate boundedness of the system. The references Beltran, Brice and Ahmed-Ali, Tarek and Benbouzid, Mohamed El Hachemi. High-order sliding-mode control of variable-speed wind turbines [J]. IEEE Transactions on Industrial electronics, 2008, 56(09), 3314-3321 and Zaafouri, Chaker and Torchani, Borhen and Sellami, Anis and Garcia, Germain. Uncertain saturated discrete-time sliding mode control for a wind turbine using a two-mass model [J]. Asian Journal of Control, 2018, 20(02), 802-818 introduce the sliding mode method to provide a robust solution, but these methods require a relatively accurate mathematical model of the wind turbine and bring non-continuous chattering phenomenon to the control signal, which can lead to a reduction in power generation efficiency.

[0007] In the field of wind power generation, the paper Chatri, Chakib and Ouassaid, Mohammed and Labbadi, Moussa and Errami, Youssef. Integral-type terminal sliding mode control approach for wind energy conversion system with uncertainties [J]. Computers and Electrical Engineering, 2022, 99: 107775. innovatively combines integral-type terminal sliding mode manifold with finite-time control, and proposes a robust control strategy. To further improve the transient and steady-state performance of the wind turbine system, the paper Wang, Jie and Bo, Didi. Adaptive fixed-time sensorless maximum power point tracking control scheme for DFIG wind energy conversion system [J]. International Journal of Electrical Power & Energy Systems, 2022, 135: 107424. proposes an improved finite-time control method, namely fixed-time control. The outstanding feature of this method is that the upper bound of the system convergence time can be pre-determined by reasonably selecting the control parameters according to the actual engineering requirements.

[0008] In summary, the existing control strategies have the following defects:

[0009] (1) Nonlinear problem: the aerodynamic torque and the pitch angle present a complex non-affine relationship, and the system parameters (such as air density, blade surface roughness) will drift with the change of environmental conditions, resulting in strong coupling and time-varying characteristics of the system dynamic response.

[0010] (2) Existing control methods are strongly dependent on the model and are prone to control signal jitter: to improve the tracking performance of the wind turbine system, researchers have proposed various robust control strategies, including sliding mode control and fixed-time control, but there are still problems such as strong dependence on the model and control signal jitter. SUMMARY

[0011] The technical problems solved by the present application are: overcoming the deficiencies of the prior art, providing a pre-defined space-time pitch angle control method for variable speed wind turbine, which divides the state space into multiple parts including buffer zone, reduces the dependence on model parameters, and weakens the chattering phenomenon by using improved barrier function, thereby improving the power generation efficiency and reducing the operating cost.

[0012] The technical scheme adopted by the present application to solve its technical problems is: the pre-defined space-time pitch angle control method for variable speed wind turbine, characterized by comprising the following steps:

[0013] Step 1: obtaining the dynamic model of variable speed wind turbine by using Betz theory;

[0014] Step 2, obtaining the pre-defined space-time approaching law with buffer zone according to the improved barrier function;

[0015] Step 3, constructing a sliding mode variable, converting the non-affine model of variable speed wind turbine into an affine model based on unknown smooth nonlinear function;

[0016] Step 4, compensating the unknown smooth nonlinear function by using neural network, and obtaining the pitch angle controller according to the neural network.

[0017] Preferably, in step 2, the pre-defined space-time approaching law is expressed as:

[0018]

[0019] Wherein,

[0020]

[0021] T, a, b, p, q, l>0, pl<1, ql>1, k2>M>0, T represents the pre-defined time to reach a specific position in space; s represents the state variable of system error, is the Gamma function, and is a positive parameter, I A (s), I B (s) and I C (s) represent the indicator function, k b represents the improved barrier function, ζ>0, which is a constant.

[0022] Preferably, in step 2, the improved barrier function k b is expressed as:

[0023]

[0024] where k2> M> 0 is a constant and ζ> 0 is a constant, s represents the state variable of system error.

[0025] Preferably, in step 3, the sliding variable is expressed as:

[0026]

[0027] After derivation, we get:

[0028]

[0029] where J t represents the total inertia of the transmission system, c is a constant, T a represents the aerodynamic torque of the wind turbine, K t represents the total damping, T g represents the equivalent electromagnetic torque, ω r represents the rotor speed, e represents the speed tracking error.

[0030] Preferably, in step 1, the dynamic model of the variable speed wind turbine is:

[0031]

[0032] where J t represents the total inertia of the transmission system, represents the total damping, T g represents the equivalent electromagnetic torque, ω r represents the rotor speed, T a represents the aerodynamic torque of the wind turbine.

[0033] Preferably, the derivative of T a of the aerodynamic torque of the wind turbine is:

[0034]

[0035] After substituting the derivative of the sliding variable, we get the affine model of the variable speed wind turbine based on unknown smooth nonlinear functions:

[0036]

[0037] where f and g are unknown smooth nonlinear functions, described as:

[0038]

[0039] where T a represents the aerodynamic torque of the wind turbine, ω r represents the rotor speed of the wind turbine, J t represents the total inertia of the transmission system, K t represents the total damping, Tg denotes the equivalent electromagnetic torque, β denotes the blade pitch angle, denotes the wind turbine rotor acceleration, c is a constant.

[0040] Preferably, in step 1, the pitch angle controller is expressed as:

[0041]

[0042] wherein, is the estimate of W, η is the learning rate, φ(Z) is the basis function of the NN, γ N is a constant, g is an unknown smooth nonlinear function, k b is an improved barrier function, I A (s), I B (s) and I C (s) denotes the indicator function,

[0043] The adaptive law is designed as:

[0044]

[0045] Preferably, in step 2, the state space is divided into three regions A, B and C, wherein:

[0046] Region A is expressed as: A = {s: 0 ≤ |s| ≤ ε-ζ};

[0047] Region B is expressed as: B = {s: ε-ζ < |s| < ε};

[0048] Region C is expressed as: C = {s: |s| > ε},

[0049] and ε > ζ > 0, ζ is a constant, ζ > 0, wherein region B is defined as a buffer region.

[0050] Compared with the prior art, the application has the beneficial effects that:

[0051] In the predefined spatiotemporal pitch angle control method of the variable-speed wind turbine generator set, the state space is divided into multiple parts including a buffer region, the dependence on model parameters is reduced, the chattering phenomenon is weakened by using an improved barrier function, and the power generation efficiency is improved and the operating cost is reduced.

[0052] In the predefined spatiotemporal pitch angle control method of the variable-speed wind turbine generator set, the unknown smooth nonlinear function in the dynamic model of the pitch angle control of the variable-speed wind turbine generator set is compensated by using the method of the neural network, and the controller and the adaptive law proposed in combination with the neural network.

[0053] ​In the predefined time-space pitch angle control method of the variable speed wind turbine generator set of the application, a predefined time-space approaching law is proposed, the state space is divided into three regions with a buffer zone, and different targets are achieved in different regions. The rapidity and robustness of the adjustment can be guaranteed.

[0054] The barrier function is improved by using tan function and tanh function, and a predefined time approaching law using system state is proposed to improve the traditional predefined time stability. BRIEF DESCRIPTION OF DRAWINGS

[0055] Figure 1 A flowchart of the predefined time-space pitch angle control method of the variable speed wind turbine generator set.

[0056] Figure 2 A wind turbine output power curve.

[0057] Figure 3 A power coefficient curve at different pitch angle values.

[0058] Figure 4 A variable speed wind turbine structure schematic diagram.

[0059] Figure 5 A state space diagram of the predefined time-space approaching law.

[0060] Figure 6 An improved barrier function curve. DETAILED DESCRIPTION

[0061] Figures 1-6 The best embodiment of the application is described below in conjunction with the accompanying drawings Figures 1-6 The application is further described.

[0062] As Figure 1 shown, a predefined time-space pitch angle control method (hereinafter referred to as pitch angle control method) of a variable speed wind turbine generator set includes the following steps:

[0063] Step 1: Obtain the dynamic model of the variable speed wind turbine using the Betz theory;

[0064] From the Betz theory, the aerodynamic power curve extracted by the wind turbine can be represented as shown in Figure 2 :

[0065]

[0066] Where ρ represents air density, v represents wind speed, R represents the radius of the wind turbine blades, C p (λ,β) represents the power coefficient of the wind turbine, β represents the blade pitch angle, and λ represents the tip speed ratio, which can be represented as:

[0067]

[0068] where ω r represents the rotor speed, R represents the blade radius of the wind turbine, and v represents the wind speed.

[0069] The aerodynamic torque T a of the wind turbine can be expressed as:

[0070]

[0071] where C p (λ, β) represents the power coefficient of the wind turbine, is a nonlinear function of the tip speed ratio λ and the pitch angle β, and can be expressed as:

[0072]

[0073] where h1 = 0.5176, h2 = 116, h3 = 0.4, h4 = 5, h5 = 21, h6 = 0.0068, and λ i represents the different tip speed ratios under different h i .

[0074] The curve of C p (λ, β) with respect to TSR(λ) under different β values is shown in the following figure. Figure 3

[0075] The structure diagram of the wind turbine rotor side and the generator side is shown in the following figure. Figure 4 The dynamic model can be expressed as:

[0076]

[0077] where J r and J g represent the inertia of the wind turbine rotor and the generator rotor, respectively; T a represents the aerodynamic torque of the wind turbine, ω r represents the wind turbine rotor speed, ω g represents the generator rotor speed, represents the wind turbine rotor acceleration, represents the generator rotor acceleration, T ls represents the low-speed shaft torque, T em represents the generator torque, T hs represents the high-speed shaft torque, K r is the external damping coefficient of the wind turbine, K g is the external damping coefficient of the generator, and n g is the relationship between them.

[0078] ​

[0079] According to equations (6)-(8), the dynamic model of the variable speed wind turbine can be described as:

[0080]

[0081] wherein, represents the total inertia of the drive train, represents the total damping, T g = n g T em represents the equivalent electromagnetic torque, ω r represents the rotor speed, T a represents the aerodynamic torque of the wind turbine.

[0082] Next, the generator output power P g is described as follows:

[0083] P g = T g ω r (10)

[0084] wherein, T g = n g T em represents the equivalent electromagnetic torque, ω r represents the rotor speed.

[0085] Step 2, obtain the pre-defined time-space approaching law with buffer zone according to the improved barrier function;

[0086] As Figure 5 shown in the present pitch angle control method, the state space is divided into three regions A, B and C, wherein:

[0087] Region A is represented as: A = {s: 0≤|s|≤ε-ζ};

[0088] Region B is represented as: B = {s: ε-ζ<|s|<ε};

[0089] Region C is represented as: C = {s: |s|>ε},

[0090] and ε>ζ>0 are constants, ε represents the boundary between region C and B, and ζ represents the boundary between region B and A.

[0091] The pre-defined time-space approaching law can be represented as:

[0092]

[0093] wherein, s is a sliding mode variable about the speed tracking error e = ω r -ω d .

[0094]

[0095]

[0096] T, a, b, p, q, l > 0, and satisfy pl < 1 < ql, and k2 > M > 0,

[0097] T represents a predefined time to reach a certain position in space, a, b, p, q, l are parameters to be designed, s (s ∈ R n ) represents the state variable of system error, is represented as:

[0098]

[0099] wherein, is the Gamma function, and is a positive definite parameter. I A (s), I B (s) and I C (s) represent the indicator function, respectively represented as:

[0100]

[0101] k b represents the improved barrier function, described as:

[0102]

[0103] wherein k2 > M > 0, which is a constant, and ζ > 0 is a very small constant. The curve of the improved barrier function is shown in Figure 6 , and further known by Figure 6 :

[0104] When the absolute value of the system state |s| is far away from s = 0, i.e. |s| ∈ C, then I C (s) = 1, I A (s) = I B (s) = 0, the predefined spatiotemporal reaching law becomes This can make the system state reach |s| = ε within a predefined time T, which guarantees the rapidity of the system state reaching |s| = ε.

[0105] Then, when the system state reaches |s| = ε, i.e. |s| ∈ B, then I B (s) = 1, I A (s) = I C (s) = 0, the predefined spatiotemporal reaching law becomes Region B is defined as a buffer zone where the reaching law has a small gain so that it can act as a buffer for the fast convergence in region C, in which the system state can approach |s| = ε - ζ at a slower speed.

[0106] Finally, when the system state reaches |s| = ε - ζ, i.e., |s| ∈ A, then I A (s) = 1, I B (s) = I C (s) = 0, the pre-defined time-space reaching law becomes From equation (13), it can be seen that k b has a unique minimum at s = 0: k b = 0, and it increases or decreases according to the distance between s and 0, specifically, k b increases as s moves away from 0 and decreases as s approaches 0. This reduces the overshoot problem and weakens the chattering phenomenon.

[0107] Step 3, construct a sliding mode variable, convert the non-affine model of the variable speed wind turbine into an affine model based on unknown smooth nonlinear functions;

[0108] Design a pre-defined time-space adaptive neural pitch angle controller using the pre-defined time-space reaching law in step 2 so that the turbine speed ω r quickly tracks its rated speed ω d , all signals of the closed-loop system are bounded, and the speed tracking error e = ω r - ω d converges to a small residual within a predefined time.

[0109] Convert the non-affine model of the variable speed wind turbine in equation (9) into an affine model, first define the sliding mode variable:

[0110]

[0111] where c is a normal number, and e represents the speed tracking error.

[0112] Differentiate equation (14) to obtain:

[0113]

[0114] where J t represents the total inertia of the transmission system, c is a normal number, T a represents the aerodynamic torque of the wind turbine, K t represents the total damping, T g represents the equivalent electromagnetic torque, ω r represents the rotor speed.

[0115] By combining (3) and (9), the aerodynamic torque T of the wind turbine can be obtained a The derivative of T with respect to β is

[0116]

[0117] Substituting (16) into (15), we have

[0118]

[0119] where f and g are unknown smooth nonlinear functions, which can be described as

[0120]

[0121] Step 4: Compensate the unknown smooth nonlinear functions f and g by using a neural network, and obtain the pitch angle controller based on the neural network.

[0122] For the controller design, the following assumptions need to be satisfied:

[0123] The unknown nonlinear component g in the controlled system (17) is available, and without loss of generality, it is assumed that there exists an unknown constant g0 such that 0 < g0 < g holds.

[0124] From (3) and (5), it can be seen that the aerodynamic torque T a is a decreasing function of β, which means that g > 0. In addition, considering the physical characteristics of the wind turbine system, the signals g and are both bounded. Therefore, the assumption is valid.

[0125] The unknown smooth nonlinear function f in (17) can be compensated by a NN in the following form:

[0126] f = W T φ(Z) + γ (18)

[0127] where W is an unknown estimate weight, Z is an input vector, γ is an approximation error defined by (19), and satisfies |γ| ≤ γ N , where γ N is a positive constant, and φ(Z) is the basis function of the NN given by

[0128]

[0129] where l1, l2, l3, and l4 are appropriate positive constants.

[0130] With the compensation of the unknown smooth nonlinear function f by the neural network, the pitch angle controller can be designed as

[0131]

[0132] wherein, is the estimation of W, designed The adaptive law of is:

[0133]

[0134] wherein, η is the learning rate, and satisfies η>0.

[0135] The stability analysis of the above obtained pitch angle controller is as follows, specifically including the following steps:

[0136] Firstly, a positive definite Lyapunov function is constructed as:

[0137]

[0138] wherein is the weighted error. The derivation of formula (22) can be obtained as:

[0139]

[0140] The adaptive law of the pitch angle controller and is brought into it, and the following can be obtained:

[0141]

[0142] Since there is a "buffer zone" in the pre-defined time-space approaching law, the system stability analysis needs to be studied in three parts:

[0143] First stage:

[0144] Firstly, when the system state is far away from the sliding mode surface s=0, i.e. |s|>ε, the system state s is located in the region C, i.e. |s|∈C, according to the properties of the indicator function, I(A)=I(B)=0, I(C)=1 can be obtained, and then the Lyapunov function is defined as:

[0145]

[0146] wherein, the definition The adaptive law of is:

[0147]

[0148] The derivation of formula (25) can be obtained as:

[0149]

[0150] wherein Integrating the above equation, we have

[0151]

[0152] We have

[0153]

[0154] where V(0) - V(T) is a constant.

[0155] Suppose there exist σ > 0 and δ > 0 such that V(t) > σ ≠ 0 for t ∈ [τ - δ, τ]. s3 Combining equation (29), we have V(t) > σ ≠ 0 for t ∈ [T - δ, T], which leads to a contradiction. Therefore, the assumption is not true, i.e.,

[0156]

[0157] If δ is a small constant, then is a positive finite term, which leads to the value of V(0) - V(T) being less than a positive lower bound, thus resulting in a contradiction. Therefore, the assumption is not true, i.e., Combining the existence and continuity of the solution, we have V(T) = 0.

[0158] As t → T, V3 is monotonically decreasing and tends to a constant, V s3 converges to 0, and V W3 is bounded. For any t ≥ T, we have V s3 = 0, V(t) = V2(t) > 0, which leads to

[0159]

[0160] Then we have

[0161]

[0162] According to equation (32) above, when i.e., we have the following inequality:

[0163]

[0164] Transforming the above inequality, we have

[0165]

[0166] Integrating both sides of the above inequality, we have

[0167]

[0168] We can find that is positive definite, then equation (35) is monotonically increasing. Therefore, the upper bound of τ can be obtained as:

[0169]

[0170] Since p - q < 0, when V s3 → 0, tends to 0, when V s3 → ∞, y → 1. Defining y as a variable change, we can calculate and its derivative can be obtained as:

[0171]

[0172] Therefore, equation (36) can be solved as:

[0173]

[0174] where B(m p ,m q ) is the beta function, which is defined as:

[0175]

[0176] From the definitions of and , and combining equation (12) and equation (37), we can obtain:

[0177]

[0178] From the above analysis, it can be obtained that the system state s can reach |s| = ε within a predefined time T.

[0179] Second stage:

[0180] When the system state s reaches |s| = ε, the system state enters the buffer zone, and the system state is located in the region B, i.e., |s| ∈ B. According to the properties of the indicator function, we can obtain I(A) = I(C) = 0, I(B) = 1. Then, define the Lyapunov function as:

[0181]

[0182] and the adaptive law of is:

[0183]

[0184] Taking the derivative of (40) gives:

[0185]

[0186] Therefore, the system is asymptotically stable in region B with the proper selection of k1>0.

[0187] The third stage:

[0188] When the system state s reaches |s| = ε - ζ, the system state is located in region A, i.e., |s| ∈ A, according to the property of the indicator function, I(B) = I(C) = 0 and I(A) = 1, and the Lyapunov function is defined as:

[0189] V1 = V s1 + V W1 (43)

[0190] And the approaching law of k is:

[0191]

[0192] The derivative of formula (43) is obtained as:

[0193]

[0194] Through the definition of k b , it can be clearly found that k b > 0. It can be obtained that By using the Lyapunov stability, it can be obtained that the system is asymptotically stable in region A.

[0195] The above is only the preferred embodiment of the present application, and is not intended to limit the other forms of the present application. Any skilled person in the art can use the disclosed technical content to make changes or modifications into equivalent embodiments. However, any simple modification, equivalent change and modification made according to the technical essence of the present application to the above embodiments without departing from the technical solution of the present application still belongs to the protection scope of the present application.

Claims

1. A method of predefined spatiotemporal pitch angle control of a variable speed wind turbine generator system, characterized by: The method comprises the following steps: Step 1: obtaining a dynamic model of a variable-speed wind turbine by using the Betz theory; Step 2: obtaining a pre-defined time-space reaching law with a buffer zone according to an improved barrier function; Step 3: constructing a sliding mode variable, converting a non-affine model of the variable-speed wind turbine into an affine model based on unknown smooth nonlinear functions; Step 4: compensating the unknown smooth nonlinear functions by using a neural network, and obtaining a pitch angle controller according to the neural network.

2. The method of pre-defined spatiotemporal pitch angle control of a variable speed wind turbine generator system as claimed in claim 1, wherein: In step 2, the pre-defined time-space reaching law is expressed as: wherein T, a, b, p, q, l > 0, pl < 1, ql > 1, k2 > M > 0, T denotes a predefined time to reach a certain position in space; s denotes a state variable of the system error, is the Gamma function, and is a positive definite parameter, I A (s), I B (s) and I C (s) denotes an indicator function, k b denotes a modified barrier function, ζ > 0, is a constant.

3. The method of pre-defined spatio-temporal pitch angle control of a variable speed wind turbine generator system as claimed in claim 2, wherein: In step 2, the improved barrier function k b is represented as: Wherein, k2>M>0 are constants, ζ>0 is a constant, and s represents a state variable of a system error.

4. The method of pre-defined spatiotemporal pitch angle control of a variable speed wind turbine generator system as claimed in claim 1, wherein: In step 3, the sliding mode variable is expressed as: After derivation, the following equation is obtained: where J t represents the total inertia of the drive train, c is a constant, T a represents the aerodynamic torque of the wind turbine, K t represents the total damping, T g represents the equivalent electromagnetic torque, ω r represents the rotor speed, e represents the speed tracking error.

5. The method of pre-defined spatio-temporal pitch angle control of a variable speed wind turbine generator system as claimed in claim 4 wherein: In step 1, the dynamic model of the variable-speed wind turbine is: where J t represents the total inertia of the drive train, represents the total damping, T g represents the equivalent electromagnetic torque, ω r represents the rotor speed, T a represents the aerodynamic torque of the wind turbine.

6. The method of pre-defined spatio-temporal pitch angle control of a variable speed wind turbine generator system as claimed in claim 5 wherein: The aerodynamic torque T of a wind turbine a The derivative is: After the derivative of the sliding mode variable is introduced, the affine model of the variable-speed wind turbine based on unknown smooth nonlinear functions is obtained: Wherein, f and g are unknown smooth nonlinear functions, and are described as: where T a represents the aerodynamic torque of the wind turbine, ω r represents the wind turbine rotor speed, J t represents the total inertia of the drive train, K t represents the total damping, T g represents the equivalent electromagnetic torque, β represents the blade pitch angle, represents the wind turbine rotor acceleration, c is a constant.

7. The method of pre-defined spatio-temporal pitch angle control of variable speed wind turbine generator system as claimed in claim 2 wherein: In step 1, the pitch angle controller is represented as: where, is the estimate of W, η is the learning rate, φ(Z) is the basis function of the NN, γ N is a positive constant, g is an unknown smooth nonlinear function, k b is the improved barrier function, I A (s), I B (s) and I C (s) represent the indicator function, Design The adaptive law is:

8. The method of pre-defined spatio-temporal pitch angle control of variable speed wind turbine generator system as claimed in claim 1 wherein: In step 2, the state space is divided into three regions A, B and C, wherein: Region A is expressed as: A={s:0≤|s|≤ε-ζ}; Region B is expressed as: B={s:ε-ζ<|s|<ε}; Region C is expressed as: C={s:|s|>ε}, And ε>ζ>0, ζ is a constant, and ζ>0, wherein region B is defined as a buffer zone.

Images