Wind power plant power coordinated distribution optimization method considering torque ripple suppression

Abstract

The invention provides a wind power plant power coordinated distribution optimization method considering torque ripple suppression, and the method comprises the following steps: calculating a corresponding unit compensation torque through the output variation of a wind turbine unit, and building a discretization equation of wind power plant power distribution based on a model predictive control (MPC) standard form; establishing a wind power plant power coordinated distribution model on the basis of tracking a wind power tracking instruction by taking minimization of torque ripple and power dispersion as optimization targets; and finally, solving the wind power plant power distribution model in a distributed cooperative manner. According to the method, the compensation torque fluctuation of the fan can be effectively reduced, and the fatigue load of the unit in the field is reduced while the operation stability of the unit is improved.

Description

Technical Field

[0001] The present invention relates to the field of wind power optimization and distribution, and in particular to a wind farm power coordinated distribution optimization method taking torque fluctuation suppression into account. Background Art

[0002] In recent years, in order to deal with the frequency problems caused by wind power grid connection and improve the frequency stability of the power system, it is required that wind farms should have a certain frequency support capability. Wind farms need to adjust their own power increment to achieve system frequency support when participating in frequency regulation. Wind farms adjust power increments through the action of each wind turbine in the field. The operating status of each unit is different. How to formulate reasonable optimization goals according to different unit conditions and achieve reasonable coordinated allocation of power increments of units in the wind farm is something that needs to be considered. At present, wind farms are usually regarded as a whole in wind power frequency regulation, lacking analysis and quantification of the operating status of units in the field. The output of each wind turbine should be coordinated and arranged under the wind farm frequency regulation command. In addition, while participating in frequency regulation, wind power also needs to maintain its own output stability to ensure that wind turbines can accurately respond to system control. Therefore, it is necessary to reasonably design a wind farm power allocation coordination optimization strategy that takes into account torque fluctuation suppression under frequency response to achieve reasonable optimization of wind power and ensure that wind power maintains the stability of unit operation while supporting frequency. Summary of the invention

[0003] Purpose of the invention: The purpose of the present invention is to provide a wind farm power coordination allocation optimization strategy taking into account torque fluctuation suppression under frequency response, which ensures the tracking accuracy of the wind farm output reference value instruction while suppressing the torque fluctuation of the unit to ensure safe and stable operation of the power grid.

[0004] Technical solution, in order to solve the above technical problems and achieve the above objectives, the present invention proposes a wind farm power coordinated allocation optimization method taking into account torque fluctuation suppression, the method comprising:

[0005] S1, calculate the corresponding compensation torque of the wind turbine according to the output change of the wind turbine, and establish the discretization equation of wind farm power distribution based on the standard form of model predictive control MPC;

[0006] S2, based on the discretized equation of wind farm power allocation, with minimization of torque fluctuation and power dispersion as the optimization goal, with the stable operation of the unit as the condition, and with tracking wind power tracking instructions as the constraint, a wind farm power coordination allocation model is established;

[0007] S3, decompose the global optimization problem into local optimization problems, solve the wind farm power allocation model in a distributed and collaborative manner, and obtain the optimal power of the wind turbines.

[0008] Furthermore, in step S1, the corresponding compensation torque is calculated by the output change of the wind turbine generator set as:

[0009]

[0010] Where ΔP w_i is the output change of the i-th fan, ΔT com_i is the compensation torque of the i-th fan, ω i is the rotor speed of the i-th fan;

[0011] The forward Euler method is used to discretize equation (1), and the formula is:

[0012] x(k+1)=x(k)+Δt·f(t,x(k)) (2)

[0013] In the formula, Δt is the sampling interval, t is time, x(k) is the system state variable at time k, t and x(k) are the horizontal and vertical coordinates of the current point respectively, and f(t,x(k)) represents the slope of the current point;

[0014] According to the above discrete method, the fan output change ΔP w i Discretized into:

[0015]

[0016] Where, T w_i is the inertia time constant of the i-th wind turbine, T is the sampling period; ΔP w_i (k) is the output change of the i-th wind turbine at time k, ΔP w_i_ref (k) is the reference value of wind power output change of the i-th wind turbine at time k, ΔP w_i (k+1) is the change in wind power output of the i-th wind turbine at time k+1;

[0017] The fan compensation torque ΔT com_i Discretized into:

[0018]

[0019] Where, T t_i is the torque action time constant of the i-th fan; ΔT com_i (k) is the compensation torque of the i-th fan at time k, ΔT com_i (k+1) is the compensation torque of the i-th fan at time k+1;

[0020] The standard form for setting up MPC is:

[0021]

[0022] In the formula, x(k+1) is the system state variable at time k+1, y(k) is the system output variable at time k, Δu(k) is the system control variable at time k, A, B, and C are the system matrix, control input matrix, and output matrix respectively;

[0023] Let x(k) be ΔP w_i (k), ΔT com_i (k) is the state vector at time k, then x(k+1) is ΔP w_i (k+1), ΔT com_i (k+1) is the state vector at time k+1; Δu(k) is ΔP w_i_ref (k) is the control variable at time k; y(k) is ΔP w_i (k), ΔT com_i (k) is the k-time output vector composed of

[0024] According to the MPC standard form, equations (3) and (4) are modified and integrated to establish the discretized state equation for wind farm power allocation:

[0025]

[0026] The corresponding coefficient matrix value of the above state equation (6) is:

[0027]

[0028] Based on the discrete state equation of wind farm power allocation, the system output vector y(k+j|k) in the future time domain starting from y(k) is predicted, j∈(1,N), N is the prediction time domain, y(k+j|k) represents the state vector at time k+j predicted at time k, and ΔP w_i (k+j|k), ΔT com_i (k+j|k), where ΔT com_i (k+j|k) represents the predicted value of the compensation torque of the i-th fan at time k+j, ΔP w_i (k+j|k) represents the predicted value of the change in the output of the i-th wind turbine at time k to time k+j.

[0029] Furthermore, in step S2, the modeling method is as follows:

[0030] Based on the discretized equation of wind farm power allocation, the expression of wind turbine torque fluctuation is obtained. The square of torque fluctuation is used as the measurement index of wind turbine torque fluctuation degree. The first objective function f1 of wind farm power allocation model is established as follows:

[0031]

[0032] In the formula, m is the number of wind turbines in the wind farm;

[0033] The second objective function f2 of the wind farm power allocation model is established to minimize the discrete degree of power variation of each unit in the wind farm. The discrete degree of power variation of the i-th wind turbine in the farm is represented by ΔP w_i_ref (k+j|k)-ΔP w_avg (k+j|k), ΔP w_i_ref (k+j|k) represents the optimal reference power change of the i-th wind turbine at time k+j obtained by optimization at time k, ΔP w_avg (k+j|k) represents the average value of the reference power change of each unit in the field at the corresponding time. It is calculated that the optimization target f2 is expressed by the variance reflecting the discrete degree of the allocated power, and the formula is as follows:

[0034]

[0035] Taking the above objectives into consideration, the optimization objective function of power allocation is established:

[0036]

[0037] In the formula, J w is the comprehensive optimization objective of distributed control, α and β are weight coefficients considering torque fluctuation and power dispersion respectively;

[0038] Constraints are imposed on the unit state quantities, including the following constraints:

[0039] (1) Constraints on the change in wind turbine output power

[0040] The output of wind turbines has limits, and the upper and lower limits of the output are constrained as follows:

[0041]

[0042] Where P w_i_min , P w_i_max are the upper and lower limits of the power for load shedding of the i-th wind turbine, which are dynamic values; P w_i (k) is the output power value of the i-th wind turbine at time k;

[0043] (2) Compensation torque constraint

[0044] The torque fluctuation that the transmission shaft of a wind turbine can bear is limited due to the limitation of its mechanical structure, which imposes constraints on the compensation torque:

[0045] ΔT com_i_min ≤ΔT com_i (k+j|k)≤ΔT com_i_max (12)

[0046] In the formula, ΔT com_i_min With ΔT com_i_max are the upper and lower limits of the compensation torque of the i-th fan;

[0047] Considering the wind farm reference output command issued after the system frequency response optimization, the sum of the wind turbine output is guaranteed to be consistent with the wind farm reference command in the power distribution. The formula is as follows:

[0048]

[0049] Where ΔP w_total is the reference power command of the wind farm.

[0050] Furthermore, in step S3, the solution method is as follows:

[0051] Augmented formula (10) Global optimization problem minJ w (x) is:

[0052]

[0053] In the formula, J w (x) refers to the objective function with variable x, I(x) is the penalty function for variable x exceeding the limit, and variable x is the optimal reference power change ΔP of the i-th wind turbine w_i_ref The set of (k+j|k);

[0054] The constraints are:

[0055]

[0056] Variable x is a local variable, and z is a global variable corresponding to x, both refer to ΔP w_i_ref The set of (k+j|k), Corresponding to the upper and lower limits of the wind turbine output power change and the compensation torque, A d z=b corresponds to the power command tracking constraint; where, x , Respectively represent the minimum and maximum values ​​of local variables, A d is the coefficient matrix, b is the wind farm reference output command ΔP w_total ; The Lagrangian function of the augmented optimization problem is:

[0057]

[0058] Where y is the Lagrange multiplier; ρ is the penalty parameter;

[0059] The optimization problem is decomposed into local optimization problems, and the wind farm power allocation model is solved in a distributed collaborative manner. The specific steps are as follows:

[0060] (1) Set the initial values ​​of the local variable x, global variable z, and Lagrange multiplier y to 0;

[0061] (2) Considering the tracking wind farm command, the global variable z is updated according to formula (17) to obtain the global variable value after l+1 iterations, that is, z [l+1] :

[0062]

[0063] In the formula, l is the number of iterations, x is [l] is the value of the local variable after the lth iteration, y [l] is the Lagrange multiplier after the lth iteration;

[0064] (3) Get z [l+1] After that, z [l+1] The i-th variable z i[l+1] , that is, the optimal reference power value of the i-th unit in the global variable, is sent to the corresponding local controller, and all z i[l+1] After the download is completed, each wind turbine solves the local optimization problem in parallel. The i-th variable in the local variable x, that is, x i , is the optimal reference power value of the i-th unit after local optimization solution. For the i-th unit, the optimization problem is J w_i (x i ), the corresponding augmented Lagrangian L i for:

[0065]

[0066] In the formula, I(x i ) is the variable x i The penalty function for exceeding the limit, y i[l] is the Lagrange multiplier of the i-th unit after the l-th iteration; x i After updating by minimizing formula (18), we get x i[l+1] , is x after the l+1th iteration i value, then x i[l+1] Set, and finally get x [l+1] ;

[0067] (4) Get z i[l+1] With x i[l+1] Then, the Lagrange multiplier y of the i-th unit is calculated in the local controller. i To update:

[0068]

[0069] In the formula, y i[l+1] is the Lagrange multiplier of the i-th unit after the l+1th iteration;

[0070] (5) Define the original residual p and the dual residual q:

[0071]

[0072] Stop the iteration when p and q meet the following conditions:

[0073]

[0074] In the formula, e pri With e dual are the primal feasible error and the dual feasible error, respectively, [l+！] ,q [l+！] They represent the original residual p and the dual residual q after the l+1th iteration respectively. BRIEF DESCRIPTION OF THE DRAWINGS

[0075] Figure 1 It is a flow chart of a method for optimizing wind farm power coordination and allocation taking torque fluctuation suppression into consideration provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0076] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0077] See also Figure 1 The present invention proposes a wind farm power coordinated allocation optimization method taking into account torque fluctuation suppression, the method comprising:

[0078] S1, calculate the corresponding compensation torque of the wind turbine according to the output change of the wind turbine, and establish the discretization equation of wind farm power distribution based on the standard form of model predictive control MPC;

[0079] S2, based on the discretized equation of wind farm power allocation, with minimization of torque fluctuation and power dispersion as the optimization goal, with the stable operation of the unit as the condition, and with tracking wind power tracking instructions as the constraint, a wind farm power coordination allocation model is established;

[0080] S3, decompose the global optimization problem into local optimization problems, solve the wind farm power allocation model in a distributed and collaborative manner, and obtain the optimal power of the wind turbines.

[0081] Furthermore, in step S1, the corresponding compensation torque is calculated by the output change of the wind turbine generator set as:

[0082]

[0083] Where ΔP w_i is the output change of the i-th fan, ΔT com_i is the compensation torque of the i-th fan, ω i is the rotor speed of the i-th fan;

[0084] The forward Euler method is used to discretize equation (1), and the formula is:

[0085] x(k+1)=x(k)+Δt·f(t,x(k)) (2)

[0086] In the formula, Δt is the sampling interval, t is time, x(k) is the system state variable at time k, t and x(k) are the horizontal and vertical coordinates of the current point respectively, and f(t,x(k)) represents the slope of the current point;

[0087] According to the above discrete method, the fan output change ΔP w i Discretized into:

[0088]

[0089] Where, T w_i is the inertia time constant of the i-th wind turbine, T is the sampling period; ΔP w_i (k) is the output change of the i-th wind turbine at time k, ΔP w_i_ref (k) is the reference value of wind power output change of the i-th wind turbine at time k, ΔP w_i (k+1) is the change in wind power output of the i-th wind turbine at time k+1;

[0090] The fan compensation torque ΔT com_i Discretized into:

[0091]

[0092] Where, T t_i is the torque action time constant of the i-th fan; ΔT com_i (k) is the compensation torque of the i-th fan at time k, ΔT com_i (k+1) is the compensation torque of the i-th fan at time k+1;

[0093] The standard form for setting up MPC is:

[0094]

[0095] In the formula, x(k+1) is the system state variable at time k+1, y(k) is the system output variable at time k, Δu(k) is the system control variable at time k, A, B, and C are the system matrix, control input matrix, and output matrix respectively;

[0096] Let x(k) be ΔP w_i (k), ΔT com_i (k) is the state vector at time k, then x(k+1) is ΔP w_i (k+1), ΔT com_i (k+1) is the state vector at time k+1; Δu(k) is ΔP w_i_ref (k) is the control variable at time k; y(k) is ΔP w_i (k), ΔT com_i (k) is the k-time output vector composed of

[0097] According to the MPC standard form, equations (3) and (4) are modified and integrated to establish the discretized state equation for wind farm power allocation:

[0098]

[0099] The corresponding coefficient matrix value of the above state equation (6) is:

[0100]

[0101] Based on the discrete state equation of wind farm power allocation, the system output vector y(k+j|k) in the future time domain starting from y(k) is predicted, j∈(1,N), N is the prediction time domain, y(k+j|k) represents the state vector at time k+j predicted at time k, and ΔP w_i (k+j|k), ΔT com_i (k+j|k), where ΔT com_i (k+j|k) represents the predicted value of the compensation torque of the i-th fan at time k+j, ΔP w_i (k+j|k) represents the predicted value of the change in the output of the i-th wind turbine at time k to time k+j.

[0102] Furthermore, in step S2, the modeling method is as follows:

[0103] Based on the discretized equation of wind farm power allocation, the expression of wind turbine torque fluctuation is obtained. The square of torque fluctuation is used as the measurement index of wind turbine torque fluctuation degree. The first objective function f1 of wind farm power allocation model is established as follows:

[0104]

[0105] In the formula, m is the number of wind turbines in the wind farm;

[0106] The second objective function f2 of the wind farm power allocation model is established to minimize the discrete degree of power variation of each unit in the wind farm. The discrete degree of power variation of the i-th wind turbine in the farm is represented by ΔP w_i_ref (k+j|k)-ΔP w_avg (k+j|k), ΔP w_i_ref (k+j|k) represents the optimal reference power change of the i-th wind turbine at time k+j obtained by optimization at time k, ΔP w_avg (k+j|k) represents the average value of the reference power change of each unit in the field at the corresponding time. It is calculated that the optimization target f2 is expressed by the variance reflecting the discrete degree of the allocated power, and the formula is as follows:

[0107]

[0108] Taking the above objectives into consideration, the optimization objective function of power allocation is established:

[0109]

[0110] In the formula, J w is the comprehensive optimization objective of distributed control, α and β are weight coefficients considering torque fluctuation and power dispersion respectively;

[0111] Constraints are imposed on the unit state quantities, including the following constraints:

[0112] (1) Constraints on the change in wind turbine output power

[0113] The output of wind turbines has limits, and the upper and lower limits of the output are constrained as follows:

[0114]

[0115] Where P w_i_min , P w_i_max are the upper and lower limits of the power for load shedding of the i-th wind turbine, which are dynamic values; P w_i (k) is the output power value of the i-th wind turbine at time k;

[0116] (2) Compensation torque constraint

[0117] The torque fluctuation that the transmission shaft of a wind turbine can bear is limited due to the limitation of its mechanical structure, which imposes constraints on the compensation torque:

[0118] ΔT com_i_min ≤ΔT com_i (k+j|k)≤ΔT com_i_max (12)

[0119] In the formula, ΔTcom_i_min With ΔT com_i_max are the upper and lower limits of the compensation torque of the i-th fan;

[0120] Considering the wind farm reference output command issued after the system frequency response optimization, the sum of the wind turbine output is guaranteed to be consistent with the wind farm reference command in the power distribution. The formula is as follows:

[0121]

[0122] Where ΔP w_total is the reference power command of the wind farm.

[0123] Furthermore, in step S3, the solution method is as follows:

[0124] Augmented formula (10) Global optimization problem minJ w (x) is:

[0125]

[0126] In the formula, J w (x) refers to the objective function with variable x, I(x) is the penalty function for variable x exceeding the limit, and variable x is the optimal reference power change ΔP of the i-th wind turbine w_i_ref The set of (k+j|k);

[0127] The constraints are:

[0128]

[0129] Variable x is a local variable, and z is a global variable corresponding to x, both refer to ΔP w_i_ref The set of (k+j|k), Corresponding to the upper and lower limits of the wind turbine output power change and the compensation torque, A d z=b corresponds to the power command tracking constraint; where, x , Respectively represent the minimum and maximum values ​​of local variables, A d is the coefficient matrix, b is the wind farm reference output command ΔP w_total ;

[0130] The Lagrangian function of the augmented optimization problem is:

[0131]

[0132] Where y is the Lagrange multiplier; ρ is the penalty parameter;

[0133] The optimization problem is decomposed into local optimization problems, and the wind farm power allocation model is solved in a distributed collaborative manner. The specific steps are as follows:

[0134] (1) Set the initial values ​​of the local variable x, global variable z, and Lagrange multiplier y to 0;

[0135] (2) Considering the tracking wind farm command, the global variable z is updated according to formula (17) to obtain the global variable value after l+1 iterations, that is, z [l+1] :

[0136]

[0137] In the formula, l is the number of iterations, x is [l] is the value of the local variable after the lth iteration, y [l] is the Lagrange multiplier after the lth iteration;

[0138] (3) Get z [l+1] After that, z [l+1] The i-th variable z i[l+1] , that is, the optimal reference power value of the i-th unit in the global variable, is sent to the corresponding local controller, and all z i[l+1] After the download is completed, each wind turbine solves the local optimization problem in parallel. The i-th variable in the local variable x, that is, x i , is the optimal reference power value of the i-th unit after local optimization solution. For the i-th unit, the optimization problem is J w_i (x i ), the corresponding augmented Lagrangian L i for:

[0139]

[0140] In the formula, I(x i ) is the variable x i The penalty function for exceeding the limit, y i[l] is the Lagrange multiplier of the i-th unit after the l-th iteration; x i After updating by minimizing formula (18), we get x i[l+1] , is x after the l+1th iteration i value, then x i[l+1] Set, and finally get x [l+1] ;

[0141] (4) Get z i[l+1] With x i[l+1] Then, the Lagrange multiplier y of the i-th unit is calculated in the local controller. i To update:

[0142]

[0143] In the formula, y i[l+1] is the Lagrange multiplier of the i-th unit after the l+1th iteration;

[0144] (5) Define the original residual p and the dual residual q:

[0145]

[0146] Stop the iteration when p and q meet the following conditions:

[0147]

[0148] In the formula, e pri With e dual are the primal feasible error and the dual feasible error, respectively, [l+！] ,q [l+！] They represent the original residual p and the dual residual q after the l+1th iteration respectively.

Claims

1. A method for optimizing wind farm power coordination and allocation taking into account torque fluctuation suppression, characterized in that: The method comprises the following steps: S1, calculate the corresponding compensation torque of the wind turbine according to the output change of the wind turbine, and establish the discretization equation of wind farm power distribution based on the standard form of model predictive control MPC; S2, based on the discretized equation of wind farm power allocation, with minimization of torque fluctuation and power dispersion as the optimization goal, with the stable operation of the unit as the condition, and with tracking wind power tracking instructions as the constraint, a wind farm power coordination allocation model is established; S3, decompose the global optimization problem into local optimization problems, solve the wind farm power allocation model in a distributed and collaborative manner, and obtain the optimal power of the wind turbines.

2. A wind farm power coordination allocation optimization method taking into account torque fluctuation suppression according to claim 1, characterized in that: In step S1, the corresponding compensation torque is calculated by the output change of the wind turbine generator set: In the formula, ΔP w_i is the output change of the i-th fan, ΔT com_i is the compensation torque of the i-th fan, ω i is the rotor speed of the i-th fan; The forward Euler method is used to discretize equation (1), and the formula is: x(k+1)=x(k)+Δt·f(t,x(k)) (2) In the formula, Δt is the sampling interval, t is time, x(k) is the system state variable at time k, t and x(k) are used as the horizontal and vertical coordinates of the current point respectively, and f(t,x(k)) represents the slope of the current point; According to the above discrete method, the fan output change ΔP w_i Discretized into: Where, T w_i is the inertia time constant of the i-th wind turbine, T is the sampling period; ΔP w_i (k) is the output change of the i-th wind turbine at time k, ΔP w_i_ref (k) is the reference value of wind power output change of the i-th wind turbine at time k, ΔP w_i (k+1) is the change in wind power output of the i-th wind turbine at time k+1; The fan compensation torque ΔT com_i Discretized into: Where, T t_i is the torque action time constant of the i-th fan; ΔT com_i (k) is the compensation torque of the i-th fan at time k, ΔT com_i (k+1) is the compensation torque of the i-th fan at time k+1; The standard form for setting up MPC is: In the formula, x(k+1) is the system state variable at time k+1, y(k) is the system output variable at time k, Δu(k) is the system control variable at time k, A, B, and C are the system matrix, control input matrix, and output matrix respectively; Let x(k) be ΔP w_i (k), ΔT com_i (k) is the state vector at time k, then x(k+1) is ΔP w_i (k+1), ΔT com_i (k+1) is the state vector at time k+1; Δu(k) is ΔP w_i_ref (k) is the control variable at time k; y(k) is ΔP w_i (k), ΔT com_i (k) is the k-time output vector composed of According to the MPC standard form, equations (3) and (4) are modified and integrated to establish the discretized state equation for wind farm power allocation: The corresponding coefficient matrix value of the above state equation (6) is: Based on the discrete state equation of wind farm power allocation, the system output vector y(k+j|k) in the future time domain starting from y(k) is predicted, j∈(1,N), N is the prediction time domain, y(k+j|k) represents the state vector at time k+j predicted at time k, and ΔP w_i (k+j|k), ΔT com_i (k+j|k), where ΔT com_i (k+j|k) represents the predicted value of the compensation torque of the i-th fan at time k+j, ΔP w_i (k+j|k) represents the predicted value of the change in the output of the i-th wind turbine at time k to time k+j.

3. A wind farm power coordination and allocation optimization method taking into account torque fluctuation suppression according to claim 2, characterized in that: The specific method in step S2 is as follows: Based on the discretized equation of wind farm power allocation, the expression of wind turbine torque fluctuation is obtained. The square of torque fluctuation is used as the measurement index of wind turbine torque fluctuation degree. The first objective function f1 of wind farm power allocation model is established as follows: In the formula, m is the number of wind turbines in the wind farm; The second objective function f2 of the wind farm power allocation model is established to minimize the discrete degree of power variation of each unit in the wind farm. The discrete degree of power variation of the i-th wind turbine in the farm is represented by ΔP w_i_ref (k+j|k)-ΔP w_avg (k+j|k), ΔP w_i_ref (k+j|k) represents the optimal reference power change of the i-th wind turbine at time k+j obtained by optimization at time k, ΔP w_avg (k+j|k) represents the average value of the reference power change of each unit in the field at the corresponding moment. The optimization target f2 is expressed by the variance reflecting the discrete degree of the allocated power, and the formula is as follows: Taking the above objectives into consideration, the optimization objective function of power allocation is established: In the formula, J w is the comprehensive optimization objective of distributed control, α and β are weight coefficients considering torque fluctuation and power dispersion respectively; Constraints are imposed on the unit state quantities, including the following constraints: (1) Constraints on the change in wind turbine output power The output of wind turbines has limits, and the upper and lower limits of the output are constrained as follows: Where P w_i_min , P w_i_max are the upper and lower limits of the power for load shedding of the i-th wind turbine, which are dynamic values; P w_i (k) is the output power value of the i-th wind turbine at time k; (2) Compensation torque constraint The torque fluctuation that the transmission shaft of a wind turbine can bear is limited due to the limitation of its mechanical structure, which imposes constraints on the compensation torque: ΔT com_i_min ≤ΔT com_i (k+j|k)≤ΔT com_i_max (12) In the formula, ΔT com_i_min With ΔT com_i_max are the upper and lower limits of the compensation torque of the i-th fan; Considering the wind farm reference output command issued after the system frequency response optimization, the sum of the wind turbine output is guaranteed to be consistent with the wind farm reference command in the power distribution. The formula is as follows: In the formula, ΔP w_total is the reference power command of the wind farm.

4. A wind farm power coordination allocation optimization method taking into account torque fluctuation suppression according to claim 3, characterized in that: In step S3, the solution method is as follows: Augmented formula (10) Global optimization problem minJ w (x) is: In the formula, J w (x) refers to the objective function with variable x, I(x) is the penalty function for variable x exceeding the limit, and variable x is the optimal reference power change ΔP of the i-th wind turbine w_i_ref The set of (k+j|k); The constraints are: Variable x is a local variable, and z is a global variable corresponding to x, both refer to ΔP w_i_ref The set of (k+j|k), Corresponding to the upper and lower limits of the wind turbine output power change and the compensation torque, A d z=b corresponds to the power command tracking constraint; in, x , Respectively represent the minimum and maximum values ​​of local variables, A d is the coefficient matrix, b is the wind farm reference output command ΔP w_total ; The Lagrangian function of the augmented optimization problem is: In the formula, y is the Lagrange multiplier and ρ is the penalty parameter; The optimization problem is decomposed into local optimization problems, and the wind farm power allocation model is solved in a distributed collaborative manner. The specific steps are as follows: (1) Set the initial values ​​of the local variable x, global variable z, and Lagrange multiplier y to 0; (2) Considering the tracking wind farm command, the global variable z is updated according to formula (17) to obtain the global variable value after l+1 iterations, that is, z [l+1] : In the formula, l is the number of iterations, x is [l] is the value of the local variable after the lth iteration, y [l] is the Lagrange multiplier after the lth iteration; (3) Get z [l+1] After that, z [l+1] The i-th variable z i[l+1] , that is, the optimal reference power value of the i-th unit in the global variable, is sent to the corresponding local controller, and all z i[l+1] After the download is completed, each wind turbine solves the local optimization problem in parallel. The i-th variable in the local variable x, that is, x i , is the optimal reference power value of the i-th unit after local optimization solution. For the i-th unit, the corresponding augmented Lagrangian L i for: In the formula, I(x i ) is the variable x i The penalty function for exceeding the limit, y i[l] is the Lagrange multiplier of the i-th unit after the l-th iteration; x i After updating by minimizing formula (18), we get x i[l+1] , is x after the l+1th iteration i Value, according to x i[l+1] The collection finally gets x [l+1] ; (4) Get z i[l+1] With x i[l+1] Then, the Lagrange multiplier y of the i-th unit is calculated in the local controller. i To update: In the formula, y i[l+1] is the Lagrange multiplier of the i-th unit after the l+1th iteration; (5) Define the original residual p and the dual residual q: Stop the iteration when p and q meet the following conditions: In the formula, e pri With e dual are the primal feasible error and the dual feasible error, respectively, [l+！] ,q [l+！] They represent the original residual p and the dual residual q after the l+1th iteration respectively.

Images