Fan wide-area operation control method for parallel optimization of fatigue load and power tracking

Abstract

The invention discloses a fan wide-area operation control method for parallel optimization of fatigue load and power tracking, and aims to realize fatigue load optimization of a tower and a shaft system and meet power control requirements in a rated operation state. The method utilizes a model predictive control architecture and comprises the following steps: firstly, establishing a model according to dynamic characteristics of a fan; secondly, linearization is carried out on steady-state operation working points of the fan, and a linearization model library is established; then, considering power tracking and load suppression performance under a single working point, and establishing a controller library; and finally, designing a soft switching parallel control strategy based on a membership function and a controller library, and adapting to the characteristic of a wide operation interval of the wind driven generator. According to the method, fan fatigue load and power tracking instability are effectively reduced in parallel.

Description

Technical Field

[0001] The present invention relates to the technical field of wind farms, and in particular to a wind turbine wide-area operation control method with parallel optimization of fatigue load and power tracking. Background Art

[0002] As a representative of clean energy, wind power has been developed and applied on a large scale. During operation, wind turbines are affected by factors such as random wind, wind shear, and external grid disturbances, which causes the shaft system and tower of wind turbine generator sets to bear more complex dynamic loads, and also increases the instability of power control. As wind turbines become larger and larger, and the capacity of large-scale wind power grid connection increases, the flexibility of key parts such as blades, towers, and shaft systems continues to increase, and the loads borne by mechanical components also increase. However, traditional control strategies cannot simultaneously guarantee the load suppression effect and stable power output performance of wind turbines under a wide range of operating conditions. If the load optimization and power tracking capabilities are insufficient, it is easy to cause extreme conditions such as shaft failure, tower deformation, and wind turbine failure and shutdown, which seriously threatens the reliability of wind turbine operation. Therefore, how to reduce wind turbine fatigue loads and power tracking instability in parallel has become a key issue to break through the shackles of large-scale wind turbine development. Summary of the invention

[0003] In order to overcome the shortcomings of the prior art, the present invention proposes a wind turbine wide-area operation control method with parallel optimization of fatigue load and power tracking to achieve fatigue load optimization of the tower and shaft system and meet the power control requirements under rated operating conditions. This method uses the model predictive control (MPC) architecture. First, a model is established based on the dynamic characteristics of the wind turbine; second, linearization is performed for the steady-state operating point of the wind turbine and a linearization model library is established; then, considering the power tracking and load suppression performance at a single operating point, a controller library is established; finally, a soft switching parallel control strategy is designed based on the membership function and the controller library to adapt to the characteristics of the wide operating range of the wind turbine.

[0004] The technical solution adopted by the present invention to solve its technical problem is:

[0005] A fan wide-area operation control method with fatigue load and power tracking parallel optimization includes the following steps:

[0006] Step 1: Establish a nonlinear characteristic model of the wind turbine, including shaft system modeling, tower modeling, generator modeling, pitch actuator modeling, and aeroelastic modeling;

[0007] Step 2: Establish a fan linear model library, use the Jacobian matrix to complete the model linearization, and introduce the gap measurement theory to establish the model library;

[0008] Step 3: Establish a controller library, consider the power tracking and load suppression performance at a single operating point, and establish a controller library based on the model predictive control architecture;

[0009] Step 4: Design a soft switching control strategy. A soft switching wide-area control strategy is designed based on the membership function and controller library.

[0010] Further, the process of step 1 is as follows:

[0011] 1.1) Aeroelastic modeling

[0012]

[0013] Here, P a is the mechanical power exerted by the wind on the fan rotor, F a is the aerodynamic thrust, ρ is the air density, R is the blade radius, C p is the power capture factor, C t is the thrust coefficient, λ is the tip speed ratio, ω r is the rotor angular velocity, v r is the effective wind speed acting on the rotor plane;

[0014] 1.2) Fan shaft system modeling

[0015] The wind turbine shaft system is modeled as follows:

[0016]

[0017] J r is the rotor equivalent moment of inertia, J g is the equivalent moment of inertia of the generator, T s is the low speed shaft torque, T g is the generator torque, ω g is the generator angular velocity, N ratio is the fan gearbox transmission ratio;

[0018] Considering the flexible deformation between the low-speed shaft and the high-speed shaft, the following mathematical model is established:

[0019]

[0020] Among them, K d is the torsional stiffness coefficient, B d is the torsional damping coefficient, is the relative angular displacement between the low-speed shaft and the high-speed shaft;

[0021] 1.3) Tower modeling

[0022] Considering the first-order front and rear bending mode of the tower, the second-order differential equation is established:

[0023]

[0024] Here, m te is the equivalent mass of the tower, x t is the displacement of the tower top, B t is the bending damping coefficient, K t is the bending stiffness coefficient;

[0025] Among them, the relevant parameters are calculated as follows:

[0026] m te =3m b +m h +m n +0.25m t (10)

[0027] B t =4πm te f n d s (11)

[0028] K t =m te (2πf n ) 2 (12)

[0029] m b is the mass of a single blade, m h is the wheel mass, m n is the cabin mass, m t is the tower mass, f n is the modal natural frequency, d s is the tower structure damping ratio;

[0030] The bending moment load of the tower is characterized as follows:

[0031]

[0032] H t is the height of the cabin from the ground;

[0033] When the wind acts on the wind turbine, the top of the tower moves accordingly, causing the effective wind speed acting on the plane of the wind turbine rotor to change relatively:

[0034]

[0035] v is the absolute wind speed in the natural coordinate system;

[0036] 1.4) Generator modeling

[0037] Consider the generator as a first-order inertial control link

[0038]

[0039] τ g is the generator inertia time constant, T g ref is the reference generator torque output by the controller, P g is the generator power, η is the generator efficiency;

[0040] 1.5) Pitch actuator modeling

[0041]

[0042] τ β is the inertia time constant of the pitch actuator, β ref is the reference pitch angle output by the controller.

[0043] Furthermore, the process of step 2 is as follows:

[0044] 2.1) Jacobian matrix construction

[0045] The state quantity x, output quantity y, control quantity u, and interference quantity d of this system are defined as follows:

[0046]

[0047] y=(ω g ,M t ,P g ,T s ) T (19)

[0048] u=(β ref ,T g ref ) T (20)

[0049] d=v r (twenty one)

[0050] Jacobian matrix construction is a common linearization method for multidimensional vector linearization. Let f(x n )=x(x n ), for f(x n ), whose Jacobian matrix is ​​calculated as follows:

[0051]

[0052] Similarly, let g(y n )=y(y n ), for g(y n ), whose Jacobian matrix is ​​calculated as follows:

[0053]

[0054] 2.2) Linearization:

[0055] Linearize the state x

[0056]

[0057] Here x| op Refers to the value of the state quantity at the operating point, Δx refers to the increment of the state quantity relative to the operating point, and the Δ below is the expression of the increment;

[0058] Specific:

[0059]

[0060]

[0061] Here, Respectively represent T a Yes r ,β,v r The value of the partial derivative of at the operating point; Respectively represent F a Yes r ,β,v r The value of the partial derivative of at the operating point;

[0062] Linearize the output y:

[0063]

[0064] Here y| op Refers to the value of the output at the operating point;

[0065] Specific:

[0066] ΔP g =η(T g Δω g +ω g ΔT g ) (37)

[0067]

[0068] The state space equation is constructed as follows:

[0069]

[0070] y=G(x,u,d) (41)

[0071] F(x,u,d) and G(x,u,d) are both functions of x,u,d;

[0072] 2.3) Constructing a model library based on gap measurement theory

[0073] In the wind speed operating range, v d Get N linear models for interval partitioning:

[0074]

[0075] v out is the cut-out wind speed of the wind turbine, v in is the cut-in wind speed;

[0076] Based on these N models, the gap threshold δ(i,j) between adjacent models is calculated:

[0077]

[0078] is the direct gap value between the ith model and the jth model, E i (s) and F i (s) represents the transfer function D of the i-th model i The numerator and denominator of (s), i, j = 1, 2, ..., N;

[0079] D(s)=E(s)F -1 (s) (45)

[0080] They satisfy the relationship:

[0081]

[0082] Get the gap matrix:

[0083]

[0084] by Select M linear models from the gap matrix as the model library for the gap threshold, C k is the magnification factor, Represents C k δ(i,j) is rounded down to 1.

[0085] Furthermore, the process of step 3 is as follows:

[0086] 3.1) Target cost function

[0087] According to the model library, multiple linear controllers are designed at the steady-state operating point of each wind turbine model to cover the entire working area of ​​the wind turbine. Considering the power tracking and fatigue load optimization effects, the cost function J is designed. m as follows:

[0088]

[0089] N c is the control interval of the controller, J m represents the cost function of the mth controller. m (k+i) represents the predicted output given by the mth controller at the ith discrete step according to the optimization algorithm, y m ref represents the reference output of the mth controller, u m (k+i) represents the control signal given by the m-th controller according to the optimization algorithm at the i-th discrete step, Δu m (k+i) represents the rate of change of the control signal, Q m , R m , S m Represent the weight adjustment matrix of the mth controller output signal, control signal, and control signal change rate. m∈(1,2,…,M);

[0090] The output u of the mth controller m By minimizing the cost function J m get;

[0091] 3.2) Cost Function Constraints

[0092] Due to actual physical limitations, the pitch action and generator speed are subject to amplitude or rate constraints. At the same time, in order to prevent the system from generating overly aggressive control inputs and outputs, the controller constraints are set as follows:

[0093]

[0094] Here, y min and max They represent the minimum and maximum limits of the system output, including generator speed, low-speed shaft torque, tower bending moment and generator power, respectively. min and u max Represent the minimum and maximum limits of the control signal, including the generator reference torque and reference pitch angle, Δu min and Δu max Respectively represent the minimum and maximum limits of the control signal change rate.

[0095] The process of step 4 is as follows:

[0096] A single controller only works at the optimal operating point of the linearized model. Although the wide-area operation control of the fan can be achieved by switching between the above controllers, the sudden switching of the controller during the control process will cause the oscillation of the control signal. To overcome the chattering, a soft switching strategy is proposed based on the membership function, as follows:

[0097]

[0098] W i (v r ) indicates the wind speed is v r The output weight of the membership function is the trapezoidal membership function W i tra (v r ) and the triangular membership function W i tri (v r ) piecewise functions composed of mixtures;

[0099] The output signal of each local controller is weighted by the membership function to obtain the global signal output u of this strategy. all :

[0100]

[0101] u i is the control signal of the ith controller.

[0102] The beneficial effects of the present invention are as follows:

[0103] 1. The method proposed in the present invention can realize smooth switching of control signals based on a model library constructed with a larger gap threshold, thereby reducing control signal jitter generated by existing wide-area control strategies.

[0104] 2. The method proposed in this invention improves the flexibility and fault tolerance of threshold selection

[0105] 3. The method proposed in the present invention can reduce the fatigue load of the wind turbine shaft system and tower at the same time.

[0106] 4. The method proposed in the present invention can achieve a more stable power generation tracking effect while reducing the fatigue load of the wind turbine. BRIEF DESCRIPTION OF THE DRAWINGS

[0107] Figure 1 A simplified block diagram of a wind turbine system model of the present invention;

[0108] Figure 2 It is a flow chart of the soft switching wide area control method of the present invention; DETAILED DESCRIPTION

[0109] The present invention is further described below in conjunction with the accompanying drawings.

[0110] Reference Figure 1 and Figure 2 , a wind turbine wide-area operation control method with fatigue load and power tracking parallel optimization, comprising the following steps:

[0111] Step 1: Establish a nonlinear characteristic model of the wind turbine, including shaft system modeling, tower modeling, generator modeling, pitch actuator modeling, and aeroelastic modeling;

[0112] like Figure 1 This is a simplified block diagram of the wind turbine system model of the present invention. The process of step 1 is as follows:

[0113] 1.1) Aeroelastic modeling

[0114] The energy captured by the blades determines the output of the wind turbine. The mathematical model between wind energy and the mechanical and electrical energy of the generator is established through aeroelastic mechanics:

[0115]

[0116] Here, P a is the mechanical power exerted by the wind on the fan rotor, F a is the aerodynamic thrust, ρ is the air density, R is the blade radius, C p is the power capture factor, C t is the thrust coefficient, λ is the tip speed ratio, ω r is the rotor angular velocity, v r is the effective wind speed acting on the rotor plane;

[0117] 1.2) Fan shaft system modeling

[0118] Due to the complexity of wind turbines, fatigue load analysis usually focuses on the basic structure of wind turbines. The transmission shaft system is identified as one of the structures prone to failure, which has a significant impact on wind turbines. The wind turbine transmission shaft system can usually be simplified by the concentrated mass method as follows: Figure 1 In the double-mass model shown, the fatigue load of the shafting is mainly generated by the unbalanced torque on both sides, and its shafting characteristics are expressed as:

[0119]

[0120] J r is the rotor equivalent moment of inertia, J g is the equivalent moment of inertia of the generator, T s is the low speed shaft torque, T g is the generator torque, ω g is the generator angular velocity, N ratio is the fan gearbox transmission ratio;

[0121] Considering the flexible deformation between the low-speed shaft and the high-speed shaft, the following mathematical model is established based on the double mass block:

[0122]

[0123] Among them, K d is the torsional stiffness coefficient, B d is the torsional damping coefficient, is the relative angular displacement between the low-speed shaft and the high-speed shaft;

[0124] 1.3) Tower modeling

[0125] The aerodynamic thrust causes the tower to deform flexibly in front and behind, which results in the displacement of the tower top. Excessive deformation leading to tower damage will cause significant economic losses. The main source of tower damage is the aerodynamic thrust F that drives the wind rotor to rotate. a , which is related to the effective wind speed v r , pitch angle β and blade tip speed ratio λ. Due to the force mechanism of the tower and Figure 1 Similar to the cantilever beam shown, the fatigue load of the tower base is usually regarded as the fatigue load of the entire tower, and the tower base bending moment is used to evaluate the fatigue load of the tower. The tower base force and bending moment characterizations are shown in (9) and (13);

[0126] Considering the first-order forward and backward bending mode of the tower, the tower is simplified into a spring-damping mass system, and the second-order differential equation is established:

[0127]

[0128] Here, m te is the equivalent mass of the tower, x t is the displacement of the tower top, B t is the bending damping coefficient, K t is the bending stiffness coefficient;

[0129] The relevant parameters are calculated as follows:

[0130] m te =3m b +m h +m n +0.25m t (10)

[0131] B t =4πm te f n d s (11)

[0132] K t =m te (2πf n ) 2 (12)

[0133] m b is the mass of a single blade, m h is the wheel mass, m n is the cabin mass, m t is the tower mass, fn is the modal natural frequency, d s is the tower structure damping ratio;

[0134] The bending moment load of the tower is characterized as follows:

[0135]

[0136] H t is the height of the cabin from the ground;

[0137] When the wind acts on the wind turbine, the top of the tower moves accordingly, causing the effective wind speed acting on the plane of the wind turbine rotor to change relatively:

[0138]

[0139] v is the absolute wind speed in the natural coordinate system;

[0140] 1.4) Generator modeling

[0141] Since the rotor response speed of the wind turbine generator set is different from the current and voltage response speed of the grid side, the rotor side has a larger inertia, the grid side has a smaller inertia, and the response speed is faster, the present invention does not involve frequency regulation, and regards the generator as a first-order inertia control link:

[0142]

[0143] τ g is the generator inertia time constant, T g ref is the reference generator torque output by the controller, P g is the generator power, η is the generator efficiency;

[0144] 1.5) Pitch actuator modeling

[0145] In practical applications, pitch change is mainly performed in the following two situations: i) the captured wind energy exceeds the power reference value; ii) the actual speed of the generator exceeds the rated speed; In the present invention, the control of the pitch change actuator is simplified to a first-order inertia link:

[0146]

[0147] τ β is the inertia time constant of the pitch actuator, β ref is the reference pitch angle output by the controller;

[0148] Step 2: Establish a fan linear model library, use the Jacobian matrix to complete the model linearization, and introduce the gap measurement theory to establish the model library; the process is as follows:

[0149] 2.1) Jacobian matrix construction

[0150] The state quantity x, output quantity y, control quantity u, and interference quantity d of this system are defined as follows:

[0151]

[0152] y=(ω g ,M t ,P g ,T s ) T (19)

[0153] u=(β ref ,T g ref ) T (20)

[0154] d=v r (twenty one)

[0155] Jacobian matrix construction is a common linearization method for multidimensional vector linearization. Let f(x n )=x(x n ), for f(x n ), whose Jacobian matrix is ​​calculated as follows:

[0156]

[0157] Similarly, let g(y n )=y(y n ), for g(y n ), whose Jacobian matrix is ​​calculated as follows:

[0158]

[0159] 2.2) Linearization:

[0160] Linearize the state x

[0161]

[0162] Here x| op Refers to the value of the state quantity at the operating point, Δx refers to the increment of the state quantity relative to the operating point, and the Δ below is the expression of the increment;

[0163] Specific:

[0164]

[0165]

[0166] Here, Respectively represent T a Yes r ,β,vr The value of the partial derivative of at the operating point; Respectively represent F a Yes r ,β,v r The value of the partial derivative of at the operating point;

[0167] Linearize the output y:

[0168]

[0169] Here y| op Refers to the value of the output at the operating point;

[0170] Specific:

[0171] ΔP g =η(T g Δω g +ω g ΔT g ) (37)

[0172]

[0173] The state space equation is constructed as follows:

[0174]

[0175] y=G(x,u,d) (41)

[0176] F(x,u,d) and G(x,u,d) are both functions of x,u,d;

[0177] In this embodiment;

[0178] F(x,u,d)=Ax+Bu+D s d (52)

[0179] G(x,u,d)=Cx (53)

[0180] A is the system state matrix, B is the control matrix, D s is the state disturbance matrix, C is the output feedthrough matrix, and the matrices are expanded as follows:

[0181]

[0182] 2.3) Constructing a model library based on gap measurement theory

[0183] As a commonly used and powerful tool, the gap measurement theory is introduced to describe the dynamic differences between linear systems with uncertainties. The model library is constructed by calculating the gap of the linearized wind turbine model. First, the wind speed is calculated with v in the wind speed operation range. dN linear models are obtained for the interval division, and each sub-area corresponds to a specific wind speed point:

[0184]

[0185] v out is the cut-out wind speed of the wind turbine, v in is the cut-in wind speed;

[0186] Based on these N models, the gap threshold δ(i,j) between adjacent models is calculated:

[0187]

[0188] is the direct gap value between the ith model and the jth model, E i (s) and F i (s) represents the transfer function D of the i-th model i The numerator and denominator of (s), i, j = 1, 2, ..., N. As shown in equation (51), for a linear system, the corresponding transfer function is calculated by normalized coprime decomposition. The closer the gap value is to 0, the more similar the dynamic responses of the two systems are; conversely, if the difference between the two models is greater, the gap value is closer to 1. Two linear systems with a close gap can be controlled by the same controller;

[0189] D(s)=E(s)F -1 (s) (45)

[0190] They satisfy the relationship:

[0191]

[0192] Get the gap matrix:

[0193]

[0194] by Select M linear models from the gap matrix as the model library for the gap threshold, C k is the amplification factor, C k Take 10, Represents C k δ(i,j) takes an integer value of 1 downward. The closer the two wind speed areas are, the smaller the gap metric of the corresponding model is. In addition, the selection of the gap metric threshold is crucial for the construction of the model library.

[0195] Step 3: Establish a controller library, considering the power tracking and load suppression performance at a single operating point, and establish a controller library based on the model predictive control architecture; the process is as follows:

[0196] 3.1) Design cost function

[0197] Based on the construction of the model library, a linear controller can be designed at the steady-state operating point of each wind turbine model. The combination of all controllers can cover the entire wind turbine operation range. Considering the power tracking and fatigue load optimization effects, the cost function J is designed. m as follows:

[0198]

[0199] N c is the prediction interval of the controller, J m represents the cost function of the mth controller, y m (k+i) represents the predicted output given by the mth controller at the ith discrete step according to the optimization algorithm, y m ref represents the reference output of the mth controller, u m (k+i) represents the control signal given by the m-th controller according to the optimization algorithm at the i-th discrete step, Δu m (k+i) represents the rate of change of the control signal. m , R m , S m Respectively represent the weight adjustment matrix of the mth controller output signal, control signal, and control signal change rate, m∈(1,2,…,M);

[0200] The output u of the mth controller m By minimizing the cost function J m get;

[0201] 3.2) Cost Function Constraints

[0202] To prevent the system from generating overly aggressive control inputs and outputs, the constraints set for the controller are as follows:

[0203]

[0204] Here, y min and max They represent the minimum and maximum limits of the system output, including generator speed, low-speed shaft torque, tower bending moment and generator power. min and u max Represent the minimum and maximum limits of the control signal, including the generator reference torque and reference pitch angle, Δu min and Δu max Respectively represent the minimum and maximum limits of the control signal change rate;

[0205] Step 4: Design a soft switching control strategy. Based on the membership function and controller library, a soft switching wide-area control strategy is designed.

[0206] Figure 2 The flow chart of the soft handover wide area control strategy of the present invention is shown;

[0207] A single controller only works at the optimal operating point of the linearized model. In previous methods, some researchers have used smaller gap values ​​to achieve approximate soft switching between different controllers to reduce the signal oscillation caused by controller switching and achieve wide-area control effects. However, the selection of gap thresholds is heavily dependent on the nonlinearity of the system and the experience of researchers. If the gap value is too large, the control signal oscillation will intensify. If the gap threshold is too small, the number of model libraries will increase exponentially, affecting the calculation efficiency. To overcome this problem, a new soft switching strategy is proposed based on the membership function, as follows:

[0208]

[0209] W i (v r ) indicates the wind speed is v r The output weight of the membership function is the trapezoidal membership function W i tra (v r ) and the triangular membership function W i tri (v r ) is a piecewise function composed of a mixture of Figure 2 The mixed membership function shown in ;

[0210] The output signal of each local controller is weighted by the membership function to obtain the global signal output u of this strategy. all :

[0211]

[0212] u i is the control signal of the ith controller.

[0213] The soft switching strategy proposed in the present invention can realize smooth switching of control signals based on a model library constructed with a relatively large gap threshold, thereby improving the flexibility and fault tolerance of threshold selection.

[0214] The above embodiments are only preferred embodiments of the present invention and are not limitations of the technical solutions of the present invention. Any technical solution that can be implemented on the basis of the above embodiments without creative work should be deemed to fall within the scope of protection of the patent of the present invention.

Claims

1. A wind turbine wide-area operation control method with fatigue load and power tracking parallel optimization, characterized in that: The method comprises the following steps: Step 1: Establish a nonlinear characteristic model of the wind turbine, including shaft system modeling, tower modeling, generator modeling, pitch actuator modeling, and aeroelastic modeling; Step 2: Establish a fan linear model library, use the Jacobian matrix to complete the model linearization, and introduce the gap measurement theory to establish the model library; Step 3: Establish a controller library, consider the power tracking and load suppression performance at a single operating point, and establish a controller library based on the model predictive control architecture; Step 4: Design a soft switching control strategy. A soft switching wide-area control strategy is designed based on the membership function and controller library.

2. The wind turbine wide-area operation control method with fatigue load and power tracking parallel optimization according to claim 1 is characterized in that: The process of step 1 is as follows: 1.1) Aeroelastic modeling Here, P a is the mechanical power exerted by the wind on the fan rotor, F a is the aerodynamic thrust, ρ is the air density, R is the blade radius, C p is the power capture factor, C t is the thrust coefficient, λ is the tip speed ratio, ω r is the rotor angular velocity, v r is the effective wind speed acting on the rotor plane; 1.2) Fan shaft system modeling The wind turbine shaft system is modeled as follows: J r is the rotor equivalent moment of inertia, J g is the equivalent moment of inertia of the generator, T s is the low speed shaft torque, T g is the generator torque, ω g is the generator angular velocity, N ratio is the fan gearbox transmission ratio; Considering the flexible deformation between the low-speed shaft and the high-speed shaft, the following mathematical model is established: Among them, K d is the torsional stiffness coefficient, B d is the torsional damping coefficient, is the relative angular displacement between the low-speed shaft and the high-speed shaft; 1.3) Tower modeling Considering the first-order front and rear bending mode of the tower, the second-order differential equation is established: Here, m te is the equivalent mass of the tower, x t is the displacement of the tower top, B t is the bending damping coefficient, K t is the bending stiffness coefficient; Among them, the relevant parameters are calculated as follows: m te =3m b +m h +m n +0.25m t (10) B t =4πm te f n d s (11) K t =m te (2πf n ) 2 (12) m b is the mass of a single blade, m h is the wheel mass, m n is the cabin mass, m t is the tower mass, f n is the modal natural frequency, d s is the tower structure damping ratio; The bending moment load of the tower is characterized as follows: H t is the height of the cabin from the ground; When the wind acts on the wind turbine, the top of the tower moves accordingly, causing the effective wind speed acting on the plane of the wind turbine rotor to change relatively: v is the absolute wind speed in the natural coordinate system; 1.4) Generator modeling Consider the generator as a first-order inertial control link τ g is the generator inertia time constant, T g ref is the reference generator torque output by the controller, p g is the generator power, η is the generator efficiency; 1.5) Pitch actuator modeling τ β is the inertia time constant of the pitch actuator, β ref is the reference pitch angle output by the controller.

3. The wind turbine wide-area operation control method with fatigue load and power tracking parallel optimization as claimed in claim 2, characterized in that: The process of step 2 is as follows: 2.1) Jacobian matrix construction The state quantity x, output quantity y, control quantity u, and interference quantity d of this system are defined as follows: y=(ω g ,M t ,P g ,T s ) T (19)u=(β ref ,T g ref ) T (20) d=v r (21) Jacobian matrix construction is a common linearization method for multidimensional vector linearization. Let f(x n )=x(x n ), for f(x n ), whose Jacobian matrix is ​​calculated as follows: Similarly, let g(y n )=y(y n ), for g(y n ), whose Jacobian matrix is ​​calculated as follows: 2.2) Linearization: Linearize the state x Here x| op Refers to the value of the state quantity at the operating point, Δx refers to the increment of the state quantity relative to the operating point, and the Δ below is the expression of the increment; Specific: Here, Respectively represent T a Yes r ,β,v r The value of the partial derivative of at the operating point; Respectively represent F a Yes r ,β,v r The value of the partial derivative of at the operating point; Linearize the output y: Here y| op Refers to the value of the output at the operating point; Specific: ΔP g =η(T g Give g +oh g ΔT g ) (37) The state space equation can be constructed as follows: y=G(x,u,d) (41)F(x,u,d),G(x,u,d) are functions of x,u,d; 2.3) Constructing a model library based on gap measurement theory In the wind speed operating range, v d Get N linear models for interval partitioning: v out is the cut-out wind speed of the wind turbine, v in is the cut-in wind speed; Based on these N models, the gap threshold δ(i,j) between adjacent models is calculated: is the direct gap value between the ith model and the jth model, E i (s) and F i (s) represents the transfer function D of the i-th model i The numerator and denominator of (s), i,j=1,2,…,N, D(s)=E(s)F -1 (s) (45) They satisfy the relationship: Get the gap matrix: by Select M linear models from the gap matrix as the model library for the gap threshold, C k is the magnification factor, Represents C k δ(i,j) is rounded down to 1.

4. The wind turbine wide-area operation control method with fatigue load and power tracking parallel optimization as claimed in claim 3 is characterized in that: The process of step 3 is as follows: 3.1) Target cost function According to the model library, multiple linear controllers are designed at the steady-state operating point of each wind turbine model to cover the entire working area of ​​the wind turbine. Considering the power tracking and fatigue load optimization effects, the cost function J is designed. m as follows: N c is the control interval of the controller, J m represents the cost function of the mth controller, y m (k+i) represents the predicted output given by the mth controller at the ith discrete step according to the optimization algorithm, y m ref represents the reference output of the mth controller, u m (k+i) represents the control signal given by the m-th controller according to the optimization algorithm at the i-th discrete step, Δu m (k+i) represents the rate of change of the control signal, Q m , R m , S m Respectively represent the weight adjustment matrix of the mth controller output signal, control signal, and control signal change rate, m∈(1,2,…,M); The output u of the mth controller m By minimizing the cost function J m get; 3.2) Cost Function Constraints Due to actual physical limitations, the pitch action and generator speed are subject to amplitude or rate constraints. At the same time, in order to prevent the system from generating overly aggressive control inputs and outputs, the controller constraints are set as follows: Here, y min and max They represent the minimum and maximum limits of the system output, including generator speed, low-speed shaft torque, tower bending moment and generator power; u min and u max Respectively represent the minimum and maximum limits of the control signal, including the generator reference torque and reference pitch angle; Δu min and Δu max Respectively represent the minimum and maximum limits of the control signal change rate.

5. The wind turbine wide-area operation control method with fatigue load and power tracking parallel optimization as claimed in claim 4, characterized in that: The process of step 4 is as follows: The design proposes a soft switching strategy based on the membership function as follows: W i (v r ) indicates the wind speed is v r The output weight of the membership function is the trapezoidal membership function W i tra (v r ) and the triangular membership function W i tri (v r ) piecewise functions composed of mixtures; The output signal of each local controller is weighted by the membership function to obtain the global signal output u of this strategy. all : u i is the control signal of the ith controller.

Images