A parameter identification method of a wind turbine electromechanical transient frequency modulation model and a medium

Abstract

The application discloses a kind of parameter identification method and medium of wind turbine electromechanical transient frequency modulation model, the method considers inertia support and realizes the mode of primary frequency modulation based on pitch angle regulation, establishes the additional frequency modulation model of wind turbine based on variable pitch load shedding, and it is equivalent to multiple input single output identification model, based on niche particle swarm optimization algorithm to input and output data optimization identification obtains controller proportion and PI parameter identification value.The wind turbine electromechanical transient model of frequency modulation established considers the characteristics and availability of measured signal, improves the engineering practicability.Secondly, multiple input double output system is identified by step optimization, realizes the splitting of system structure, converts the problem into the parameter identification of double input single output system, avoids the identification difficulty caused by structure coupling.At the same time, due to the efficient operation of niche particle swarm optimization algorithm, the identification error of control parameter is less than 10%, and the identification accuracy is higher.

Description

Technical Field

[0001] This invention relates to wind turbine generator sets, specifically to a parameter identification method and medium for an electromechanical transient frequency regulation model of a wind turbine generator set. It is applicable to the identification of parameters of the electromechanical transient frequency regulation model of a wind turbine generator set, thereby obtaining accurate control parameters of the electromechanical transient frequency regulation model, providing a model basis for frequency stability analysis of large-scale wind power integration into the power system, and is of great significance for improving the simulation accuracy of the electromechanical transient model of a wind turbine generator set. Background Technology

[0002] In recent years, wind power generation has developed rapidly, with installed capacity increasing quickly, and the wind power penetration rate in the power grid will continue to rise in the future. At the same time, the problem of safe and stable operation of the power grid under the background of high proportion of wind power grid connection is becoming increasingly serious. At present, power grid analysis mainly relies on digital simulation or hybrid simulation. Accurate models and parameters are necessary conditions for obtaining correct simulation results. Therefore, modeling wind turbine controllers and identifying their parameters is of great research significance.

[0003] Currently, wind power primary frequency regulation methods mainly include utilizing rotor kinetic energy, load shedding, and external energy storage. For methods utilizing rotor kinetic energy, the rotor's kinetic energy is limited, and the rotational speed cannot be maintained in a deceleration or acceleration state for extended periods, resulting in relatively weak frequency regulation capabilities. For load shedding, using rotor overspeed backup will cause the unit to operate at a suboptimal power point, and the rotor overspeed control range is limited, only applicable to low to medium wind speeds. Energy storage systems offer rapid response and flexible control, making them a high-performance frequency regulation resource, but they are limited by high investment and operation / maintenance costs.

[0004] This invention establishes a transfer function model for wind turbines that considers inertia support and frequency regulation response based on pitch angle adjustment. An identification model for measured signals such as power and pitch angle is established using the bilinear transform method. Based on the niche particle swarm optimization algorithm, the control parameters and PI regulator parameters of the additional frequency control and pitch angle control in the wind turbine frequency regulation model are identified. Summary of the Invention

[0005] This invention addresses the transient frequency regulation model of wind turbine electromechanical systems. Considering inertia support and primary frequency regulation based on pitch angle adjustment, it establishes an additional frequency regulation model for wind turbines based on pitch reduction. Furthermore, it designs a parameter identification method for the wind turbine frequency regulation model based on measured data of wind turbine inertia (power, pitch angle, etc.) and primary frequency regulation. This method establishes a multi-input single-output equivalent model of the frequency regulation model and uses a niche particle swarm optimization algorithm to optimize and identify the input and output data, obtaining the identified values ​​of the controller proportional and PI parameters. This invention introduces the principle, implementation steps, and related parameters of this identification method and discloses the identification of control parameters for the transient frequency regulation model of wind turbine electromechanical systems based on this method.

[0006] The present invention provides a parameter identification method for a wind turbine electromechanical transient frequency regulation model, comprising:

[0007] Apply a grid frequency disturbance excitation signal to a normally operating wind turbine and collect the input and output variable data of the frequency regulation control system that applied the excitation signal;

[0008] The niche particle swarm optimization algorithm is used to optimize the established frequency regulation electromechanical transient model of the wind turbine based on the input and output data. The optimization result is the frequency regulation control parameter identification result.

[0009] Preferably, the frequency regulation electromechanical transient model of the wind turbine includes an additional frequency control module and a pitch angle control module, wherein...

[0010] The input and output variables of the wind turbine frequency regulation electromechanical transient model include the input quantities u1, u2 and the output quantity y1 of the additional frequency control module, and the input quantities u3, u4 and the output quantity y2 of the pitch angle control module.

[0011]

[0012]

[0013] Δf represents the frequency variation of the power system; P AGC The active power control command value issued by the wind farm's AGC; P S P cmd ω r and ω n These are the stator output active power, the stator active power reference value generated by the additional droop control, the rotor speed, and the rated speed, respectively. cmd This is a reference value for the pitch angle.

[0014]

[0015]

[0016] As a preferred option, the optimization objective functions for minimizing the fitting error of the additional frequency control module and the pitch angle control module based on the dual-input single-output discrete mathematical model are Q1 and Q2, respectively, and their expressions are shown in the following formulas:

[0017]

[0018] Where y′1(k) and y′2(k) are the discretized measurement outputs. The information vector is composed of the input variables u1(k), u1(k-1), u2(k) and u3(k), u3(k-1), u4(k), u4(k-1), respectively. Let be a parameter vector, which consists of discrete-domain parameters to be identified. and constitute.

[0019] As a preferred option, the discrete-domain parameters to be identified Let a1, a2, b1 be the true values ​​of the discrete-domain mathematical model and their corresponding identification values ​​in the identification model. The discrete-domain parameters to be identified are... Let c1, c2, d1, and d2 be the identified values ​​of the true values ​​of the discrete-domain mathematical model in the identification model. The true values ​​of the discrete-domain mathematical model can be calculated using the following formula:

[0020]

[0021] Where k p k d These are the droop parameter for the additional primary frequency control and the proportional parameter for the additional inertia control, k, respectively. pp k ip k pc and k ic These are the proportional and integral parameters for pitch angle control and the proportional and integral parameters for pitch angle compensation control, respectively.

[0022] Preferably, in the frequency control module of the wind turbine frequency regulation electromechanical transient model, the primary frequency regulation droop control responding to the frequency change generates a droop active power reference value according to the droop curve, and the inertia control responding to the frequency change rate generates an inertia active power reference value proportionally. The two are then superimposed on the received AGC command.

[0023] Preferably, in the wind turbine frequency regulation electromechanical transient model pitch angle control module, the pitch angle control loop takes the deviation between the wind turbine rotor speed and the rated speed as input, and generates a pitch angle command through a PI controller. At the same time, the pitch angle compensation loop takes the deviation between the current output active power and the power reference value generated by the additional primary frequency regulation droop control loop as input, and generates a pitch angle compensation command through a PI controller. The two are superimposed to obtain the total pitch angle reference value, which is then sent to the pitch angle mechanical link.

[0024] As a preferred option, the niche particle swarm optimization algorithm is a method for identifying and optimizing control parameters. The optimization process is divided into two stages: particle swarm update and niche swarm update. The particle swarm update equation and the niche swarm radius calculation equation are shown below:

[0025]

[0026] In the formula, ω is the inertia weight. Let be the historical best solution found from the i-th particle up to the k-th generation. This is the optimal solution found so far in the entire particle swarm. These are the current position and velocity of the i-th particle, respectively; c1 and c2 are called acceleration factors; and r1 and r2 are random numbers between [0,1]. Sj,g , They are respectively small life mirror group S j The optimal particle and all other particles in the system.

[0027] As a preferred method, niche population renewal follows the following criteria:

[0028] Criterion 1: If particle Z i Entering the Xiaoshengjing group S j Within the area, i.e. Then this particle will be S j accept.

[0029] Criterion 2: If two small-scale mirror groups S j S k Intersecting regions, i.e. The two small groups of mirror images merged.

[0030] As a preferred option, the optimization process specifically includes:

[0031] Step 1: Set algorithm parameters and initialize the particle swarm. Set the population size to N, the maximum number of iterations to M, and k... p k d k pp k ip k pc and k ic For the location parameters of an individual.

[0032] Step 2: Initialize the population and randomly generate the location of each individual;

[0033] Step 3: Calculate the fitness of all solutions using the formulas for the optimization objective functions Q1 and Q2;

[0034] Step 4: Construct a circular group of small life mirrors centered on the particle with the lowest fitness, with the radius being the distance to its nearest particle;

[0035] Step 5: Using the method of finding the optimal value for the population as the optimal value for the niche population, update the particle position and velocity according to the particle population update equation;

[0036] Step 6: Update the niche population according to the niche population update criteria;

[0037] Step 7: Update the optimal fitness and the global optimal solution;

[0038] Step 8: If Q is satisfied minIf Qmin is less than or reaches the maximum number of iterations, the iteration is terminated and the optimal solution is output as the identification result. Otherwise, repeat steps 3-7 until Qmin is less than or reaches the maximum number of iterations, then the iteration is terminated.

[0039] A non-transitory computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the parameter identification method.

[0040] Compared with existing technologies, this invention has the following advantages and effects: Compared with methods such as utilizing rotor kinetic energy, load reduction for backup, and external energy storage, variable pitch control has a certain reserve capacity, strong adjustment capability, and a large adjustment range, enabling power control under all wind speeds. The established electromechanical transient frequency modulation model considers the characteristics and availability of measured signals, improving its engineering practicality. Step-by-step optimization identification of the multi-input dual-output system achieves system structure decomposition, transforming the problem into parameter identification of a dual-input single-output system, avoiding the identification difficulties caused by structural coupling. The niche particle swarm optimization algorithm used is simple to operate and can reasonably search for the global optimum, providing strong support for the consistency of parameter identification. Attached Figure Description

[0041] Figure 1 This is the control block diagram for the electromechanical transient frequency regulation model of a wind turbine generator.

[0042] Figure 2 This is an equivalent two-input single-output system model.

[0043] Figure 3 This is a flowchart of the identification algorithm based on the niche particle swarm optimization algorithm.

[0044] Figure 4 This is a single-machine-grid electromechanical transient simulation model.

[0045] Figure 5 This shows the changes in the frequency excitation signal.

[0046] Figure 6a The additional power control identification parameter model based on simulation data is compared with the output results of the original system.

[0047] Figure 6b To compare the output results of the pitch angle control identification parameter model based on simulation data with those of the original system.

[0048] Figure 7a This shows the measured frequency changes of the input and output quantities under frequency modulation conditions.

[0049] Figure 7b This shows the changes in active power of the measured input and output quantities under frequency modulation conditions.

[0050] Figure 7cThis shows the measured input and output quantities, namely the pitch angle and rotor speed changes, under frequency modulation conditions.

[0051] Figure 8a This shows the frequency variation of the measured input and output quantities under inertial response conditions.

[0052] Figure 8b This represents the changes in active power of the measured input and output quantities under inertial response conditions.

[0053] Figure 8c This represents the measured input and output quantities, namely the pitch angle and rotor speed changes, under inertial response conditions.

[0054] Figure 9a This study compares the output results of the droop control identification parameter model in the additional power control based on measured data with those of the original measured data.

[0055] Figure 9b This study compares the output results of the inertia control identification parameter model in the additional power control based on measured data with those of the original measured data.

[0056] Figure 9c This is to compare the output results of the pitch angle control parameter identification model based on measured data with the original measured data. Detailed Implementation

[0057] The following section introduces a practical modeling method for a wind turbine electromechanical transient frequency regulation model, and analyzes and calculates it with specific examples.

[0058] The theoretical basis and methods involved in this invention will be introduced in turn below.

[0059] First, the relevant principles of this invention will be introduced.

[0060] 1. Principle of Niche Particle Swarm Optimization Algorithm

[0061] Particle Swarm Optimization (PSO) is a biomimetic algorithm that simulates flock flight of birds. It boasts advantages such as a small number of individuals, computational simplicity, and robustness, exhibiting fast convergence and strong versatility. It has achieved excellent results in various multidimensional continuous space optimization problems. To avoid premature convergence in PSO and increase the diversity of the particle population to improve convergence speed, a niche-sharing mechanism is introduced. The main idea is based on the principle of "birds of a feather flock together," reflecting how, in the evolutionary process of nature, various organisms survive in specific environments. Within the same species, there are superior organisms, leading to competition among them, while different species exchange information.

[0062] The niche particle swarm optimization algorithm is mainly divided into two stages. First, the niche technique finds the niche group of each particle based on the distance between particles. Then, the particle swarm optimization algorithm is used to update the velocity and position in each niche group. The optimal value of the particle swarm only works in that niche group.

[0063] II. The specific methods for applying the above principles are described below, including:

[0064] Step 1: Determine the structure of the frequency regulation electromechanical transient model of the wind turbine generator, determine the input and output variables of the controller, establish a multi-input single-output discrete mathematical model, and obtain the optimization objective function with the minimum fitting error.

[0065] Step 2: Apply system frequency input excitation to the normally operating controller and collect the input and output variable data of the control system to which the excitation signal is applied.

[0066] Step 3: Use the niche particle swarm optimization algorithm to find the optimal objective function based on the input and output data. The optimization result is the converter control parameter identification result.

[0067] Step 3.1, the method for obtaining the optimization objective function is as follows:

[0068] Figure 1 The diagram shows the control block diagram of the frequency regulation electromechanical transient model of a wind turbine, including an additional frequency control loop and a pitch angle control loop. In the diagram, k... p k d These are the droop parameter for the additional frequency modulation control and the proportional parameter for the additional inertia control, k, respectively. pp k ip k pc and k ic These represent the proportional and integral parameters for pitch angle control and the proportional and integral parameters for pitch angle compensation control, respectively. Δf represents the power system frequency variation; P AGC P is the active power command value for wind farm control. S P cmd ω r and ω n These are the stator output active power, the stator active power reference value generated by the additional droop control, the rotor speed, and the rated speed, respectively. cmd This is a reference value for the pitch angle.

[0069] To identify the controller parameters in the electromechanical transient frequency modulation model, the input and output quantities must first be defined. A step-by-step identification method is adopted. First, the additional power control model is considered, and two control parameters in the model are identified. The frequency modulation command is then calculated based on these parameters. Next, the four PI parameters in the pitch angle control and pitch angle compensation control are identified by substituting them into the pitch angle control model. At this point, both the additional power control model and the pitch angle control model can be equivalently represented as... Figure 2The dual-input single-output system shown here, where u1, u2 and y1 and u3, u4 and y2 are the input and output quantities of the additional power control model and the pitch angle control model, respectively.

[0070] Combination Figure 1 and Figure 2 It is easy to conclude that:

[0071]

[0072] Figure 1 The output power equation of the frequency modulation load reduction model is:

[0073]

[0074] By equating the converter to a controlled power source, we can write:

[0075] P S =P ref (3)

[0076] Substituting equation (3) into equation (2) and simplifying, we get

[0077] P S =sk d Δf+k p Δf+P AGC (4)

[0078] Similarly, the reference value for the output pitch angle in pitch angle control can be written as follows:

[0079]

[0080] Combination Figure 1 From equations (4) and (5), we can obtain the expression for the output of the frequency modulation electromechanical transient model as follows:

[0081]

[0082] In equation (6), Δf represents the frequency variation of the power system; P AGC The active power control command value issued by the wind farm's AGC; P S P cmd ω r and ω n These are the stator output active power, the stator active power reference value generated by the additional droop control, the rotor speed, and the rated speed, respectively. By querying motor parameters and measuring the stator three-phase voltage and current, as well as the speed, the outputs y1 and y2 can be obtained based on this equivalent identification mathematical model.

[0083] For the additional power control model, its transfer function expression can be written as:

[0084] Y1(s)=k pU1(s)+sk d U1(s)+U2(s) (7)

[0085] To identify the parameters of the multi-input single-output system shown in equation (7), it is necessary to discretize it. Since the Tustin transform method can guarantee that the stability of the model remains unchanged before and after discretization, and does not change its steady-state gain, this paper uses the Tustin transform method for discretization. Substituting into the equation yields

[0086] y1(k)+y1(k-1)=a1u1(k)+a2u1(k-1)+b1u2(k) (8)

[0087] in

[0088]

[0089] In the formula, c represents the Tustin transform coefficient, which satisfies c = 2 / T with respect to the sampling time T. Assume...

[0090] If y′1(k)=y1(k)+y1(k-1), then equation (8) can be rewritten as:

[0091] y′1(k)=a1u1(k)+a2u1(k-1)+b1u2(k) (10)

[0092] Equation (10) is the mathematical model of the equivalent identification system with additional power control, dual inputs, and single output. The optimization objective function can then be constructed based on the principle of minimizing the fitting error:

[0093]

[0094] For the pitch angle control model, its transfer function expression can be written as:

[0095]

[0096] After discretization, we can obtain

[0097] y2(k)-y2(k-1)=c1u3(k)+c2u3(k-1)+d1u4(k)+d2u4(k-1) (13)

[0098] in

[0099]

[0100] Assuming y'2(k) = y2(k) - y2(k-1), then equation (13) can be rewritten as:

[0101] y'2(k)=c1u3(k)+c2u3(k-1)+d1u4(k)+d2u4(k-1) (15)

[0102] Equation (15) is the mathematical model of the equivalent identification system for dual-input single-output propeller pitch angle control. The optimization objective function can then be constructed based on the principle of minimizing fitting error:

[0103]

[0104] Step 3.2, the method for acquiring the controller input and output signals is as follows:

[0105] Considering that the controller activates because the secondary-side measurement circuit detects changes in the primary-side electrical quantities, and the controller's response is triggered only after the measurement signal is sent to the controller, this invention applies system frequency input excitation to a normally operating wind turbine, collects the output variable data of the control system upon which the excitation signal is applied, such as pitch angle and stator three-phase voltage and current, and obtains the output quantities required for the identification model after data processing; simultaneously, it records input variable data such as system frequency and rotor speed.

[0106] Step 3.3, the niche particle swarm optimization algorithm optimization process is as follows:

[0107] The method of using niche particle swarm optimization (NPSO) to identify control parameters is divided into two stages: particle swarm update and niche swarm update. The particle swarm update equation is shown in Equation (17), and the niche swarm radius calculation equation is shown in Equation (18).

[0108]

[0109]

[0110] ω is the inertia weight. Let be the historical best solution found from the i-th particle up to the k-th generation. This is the optimal solution found so far in the entire particle swarm. These are the current position and flight speed of the i-th particle, respectively. c1 and c2 are called acceleration factors, and r1 and r2 are random numbers between [0,1]. They are respectively small life mirror group S j The optimal particle and all other particles in the array. For the small student mirror group S j The radius.

[0111] Microhabitat population regeneration involves two important steps:

[0112] 1) If particle Z i Entering the Xiaoshengjing group Sj Within the area, i.e. Then this particle will be S j accept.

[0113] 2) If two small mirror groups S j S k Intersecting regions, i.e. The two small groups of mirror images merged.

[0114] like Figure 3 The specific process of optimization identification based on niche particle swarm optimization algorithm is as follows:

[0115] (1) Set the algorithm parameters and initialize the particle swarm.

[0116] (2) Calculate the fitness value of each particle in the particle swarm. Construct a circular small swarm with the particle with the lowest fitness as the center. The radius of the swarm is the distance to its nearest particle.

[0117] (3) Update all particles according to the particle swarm position and velocity update formula (13), where the swarm optimal value is the optimal value of the niche swarm, and no longer the optimal value of the entire swarm.

[0118] (4) Update the niche population. For sub-particles that have not formed a niche population, if they satisfy step 1), they are accepted by this niche population. If they do not satisfy the condition, calculate the change in fitness before and after the update. If the change is small, generate a niche population centered on this particle. For particles that are already in a niche population, update the radius of their respective niche population.

[0119] (5) If two small mirror groups satisfy step 2), then the two small mirror groups are merged.

[0120] (6) Return to step (2) until the maximum number of iterations is reached.

[0121] III. The following is a specific example illustrating the above method, with the steps as follows:

[0122] Step 1: First, optimize and identify the simulation data of the frequency regulation electromechanical transient model of the wind turbine. For example... Figure 4 As shown, a single-machine-grid electromechanical transient simulation model was built on the MATLAB / SIMULINK platform. The wind turbine component includes a wind power model, a dual-mass rotor model, a converter model, and a wind turbine frequency regulation electromechanical transient model. A ramp function frequency curve was used as the excitation signal for the wind turbine frequency regulation electromechanical transient model, and its excitation signal variation is shown in the figure. Figure 5As shown, the simulation was run and input / output data was collected. Finally, the niche particle swarm optimization algorithm was used for optimization identification. The results are shown in Table 1. At the same time, the controller parameters of the frequency regulation system were set to the identification results in Table 1. The active power output of the identified parameter model and the active power output of the original system were compared as shown in Figure 6(a), and the pitch angle change of the identified parameter model and the pitch angle change of the original system were compared as shown in Figure 6(b). It can be seen that the results in Table 1 show that the relative errors of each parameter are very small. The active power output curve and pitch angle change curve of the identified parameter model in Figure 6 are in good agreement with the original system, indicating that the identification error of the parameter identification method in this paper meets the accuracy requirements, and showing that the parameter identification method of the wind turbine frequency regulation electromechanical transient model based on the niche particle swarm optimization algorithm is feasible.

[0123] Table 1. Parameter identification results of the transient model of frequency regulation electromechanical system for wind turbines based on simulation data.

[0124]

[0125] Step 2: Parameter optimization and identification of the electromechanical transient frequency regulation model based on the measured data of the wind turbine. First, the three-phase stator voltage and current signals of the wind turbine are processed to obtain the input and output variables required for parameter identification. The processed input and output variables are shown in Figures 7 and 8. The niche particle swarm optimization algorithm is used for parameter optimization and identification. The results are shown in Table 2. To verify that the parameter identification error in Table 2 meets the accuracy requirements, the system controller parameters are set to the identification results in Table 2. The active power output of the identified parameter model is compared with the measured active power output, as shown in Figures 9(a) and 9(b). The pitch angle change of the identified parameter model is compared with the measured pitch angle change, as shown in Figure 9(c). It can be seen that the active power output curve and pitch angle change curve of the identified parameter model are in good agreement with the measured data, indicating that the measured modeling method based on the wind turbine frequency regulation electromechanical transient model is effective.

[0126] Table 2. Parameter identification results of the transient model of frequency-regulating wind turbine electromechanical system based on measured data.

[0127]

[0128]

[0129] The device embodiments described above are merely illustrative. The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the modules can be selected to achieve the purpose of this embodiment according to actual needs. Those skilled in the art can understand and implement this without any creative effort.

[0130] Through the above description of the embodiments, those skilled in the art can clearly understand that each embodiment can be implemented by means of software plus necessary general-purpose hardware platforms, and of course, it can also be implemented by hardware. Based on this understanding, the above technical solutions, in essence or the part that contributes to the prior art, can be embodied in the form of a software product. This computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, magnetic disk, optical disk, etc., and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute the methods described in the various embodiments or some parts of the embodiments.

[0131] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, and not to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features; and these modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for parameter identification of a wind turbine electromechanical transient frequency regulation model, characterized in that, include: Apply a grid frequency disturbance excitation signal to a normally operating wind turbine and collect the input and output variable data of the frequency regulation control system that applied the excitation signal; The niche particle swarm optimization algorithm is used to optimize the established frequency regulation electromechanical transient model of the wind turbine based on the input and output data. The optimization result is the frequency regulation control parameter identification result. The frequency regulation electromechanical transient model of a wind turbine includes an additional frequency control module and a pitch angle control module, in which... The input and output variables of the wind turbine frequency regulation electromechanical transient model include the input quantities of the additional frequency control module. u 1. u 2 and output quantity y 1, and the input quantity of the pitch angle control module. u 3. u 4 and output y 2, Δ f For power system frequency variations; P AGC The active power control command value issued by the wind farm's AGC; P S , P cmd , ω r and ω n These are the stator output active power, the stator active power reference value generated by the additional droop control, the rotor speed, and the rated speed, respectively. β cmd This is a reference value for the pitch angle; The optimization objective function for minimizing the fitting error of the additional frequency control module and the pitch angle control module, based on a dual-input single-output discrete mathematical model, is: Q 1. Q 2, its expression is shown in the following formula: , in, , To discretize the measurement output, , The information vector consists of the input variables. u 1( k ), u 1( k -1) u 2( k )and u 3( k ), u 3( k -1) u 4( k ), u 4( k -1) Composition, , Let be a parameter vector, which consists of discrete-domain parameters to be identified. , , and , , , constitute; Discrete domain parameters to be identified , , For the true value of the discrete domain mathematical model , , The identified value in the identification model, the discrete-domain parameter to be identified. , , , For the true value of the discrete domain mathematical model , , , The identified value in the identification model, the true value of the discrete-domain mathematical model, can be calculated according to the following formula: in k p , k d These are the droop parameter for the additional primary frequency control and the proportional parameter for the additional inertia control, respectively. k pp , k ip , k pc and k ic These are the proportional and integral parameters for pitch angle control, and the proportional and integral parameters for pitch angle compensation control, respectively. c These are the Tustin transform coefficients.

2. The parameter identification method for a wind turbine electromechanical transient frequency regulation model according to claim 1, characterized in that, In the frequency control module of the wind turbine frequency regulation electromechanical transient model, the primary frequency regulation droop control responding to the frequency change generates a droop active power reference value according to the droop curve, and the inertia control responding to the frequency change rate generates an inertia active power reference value proportionally. The two are then superimposed on the received AGC command.

3. The parameter identification method for a wind turbine electromechanical transient frequency regulation model according to claim 1, characterized in that, In the pitch angle control module of the frequency regulation electromechanical transient model of the wind turbine, the pitch angle control loop takes the deviation between the wind turbine rotor speed and the rated speed as input, and generates a pitch angle command through the PI controller. At the same time, the pitch angle compensation loop takes the deviation between the current output active power and the power reference value generated by the additional primary frequency regulation droop control loop as input, and generates a pitch angle compensation command through the PI controller. The two are superimposed to obtain the total pitch angle reference value, which is then sent to the pitch angle mechanical link.

4. The parameter identification method for a wind turbine electromechanical transient frequency regulation model according to claim 1, characterized in that, The niche particle swarm optimization process is a method for identifying and optimizing control parameters using the niche particle swarm optimization algorithm. The optimization process is divided into two stages: particle swarm update and niche swarm update. The particle swarm update equation and the niche swarm radius calculation equation are shown below: In the formula ω For inertial weights, Let be the historical best solution found from the i-th particle up to the k-th generation. This is the optimal solution found so far in the entire particle swarm. , These are the current position and velocity of the i-th particle, respectively. c 1, c 2 is called the acceleration factor. r 1, r 2 is a random number between [0, 1]; , They are respectively the Xiaoshengjing group The optimal particle and all other particles in the array. For the Xiaoshengjing group The radius.

5. The parameter identification method for a wind turbine electromechanical transient frequency regulation model according to claim 1, characterized in that, Microhabitat population regeneration follows these criteria: Criterion 1: If the particle Entering the Xiaoshengjing Group Within the area, i.e. Then this particle will be accept; Criterion 2: If two small groups of mirrors , Intersecting regions, i.e. The two small groups of mirrors merged.

6. The parameter identification method for a wind turbine electromechanical transient frequency regulation model according to claim 1, characterized in that, The optimization process specifically includes Step 1: Set the algorithm parameters and initialize the particle swarm. Set the population size to N individuals and the maximum number of iterations to M. k p , k d , k pp , k ip , k pc and k ic For the individual's positional parameters; Step 2: Initialize the population and randomly generate the location of each individual; Step 3: Optimize according to the objective function. Q 1. Q Formula 2 calculates the fitness of all solutions; Step 4: Construct a circular group of small life mirrors centered on the particle with the lowest fitness, with the radius being the distance to its nearest particle; Step 5: Using the method of finding the optimal value for the population as the optimal value for the niche population, update the particle position and velocity according to the particle population update equation; Step 6: Update the niche population according to the niche population update criteria; Step 7: Update the optimal fitness and the global optimal solution; Step 8: If satisfied Q min < If the maximum number of iterations is reached, the iteration is terminated and the optimal solution is output as the identification result. Otherwise, steps 3-7 are repeated until Qmin is less than or the maximum number of iterations is reached, at which point the iteration is terminated.

7. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements the parameter identification method as described in any one of claims 1 to 6.

Images