CN116816597A - Control method and system of wind turbine generator, electronic equipment and storage medium - Google Patents

Abstract

The present disclosure provides a control method, system, electronic device and storage medium for a wind turbine. The control method includes: obtaining the inflow wind speed of the wind turbine in at least one predictive control period within the current control cycle; obtaining the operating status data of the wind turbine in the predictive control cycle based on the incoming wind speed; based on the operating status data and the generator The preset cost function generates a target control sequence; in response to entering the next control cycle, the generator torque is controlled according to the target torque value in the target control sequence. Based on the Pareto theory, this disclosure proposes a preset cost function that can collaboratively optimize the tower load of the wind turbine and the wind energy capture power, effectively reduce the lateral structural load of the tower of the wind turbine, and improve the generator. It has important engineering application value to improve the power output quality of large-scale wind turbines and improve the reliability and energy efficiency of wind turbine operations.

Description

Control methods, systems, electronic devices and storage media for wind turbines

Technical field

The present disclosure relates to the technical field of wind turbine control, and in particular to a wind turbine control method, system, electronic equipment and storage medium.

Background technique

A wind turbine is a device that converts wind energy into electrical energy and consists of multiple components. It mainly includes wind wheel (blade), generator, converter, tower and control system. The wind wheel rotates driven by the wind, activating the generator to generate electrical energy, and transmits the electrical energy to the power grid or other electrical equipment through the power transmission system.

With the increasing demand for wind energy, the rated power of wind turbines and the size of key support structures have been designed to be large-scale, resulting in obvious flexibility characteristics of the wind turbine structure.

In addition, large-scale wind turbines are prone to elastic deformation under the influence of complex aerodynamic loads, which increases the load on the tower and affects the stability and service life of the wind turbine operation.

Currently, the structural damping of the tower can be adjusted by controlling the blade pitch angle and generator torque. For example, in the process of controlling the blade pitch angle, an active damping control loop is added to reduce oscillation, achieving the goal without adding additional structures. In this case, reduce the load on the tower. However, this method mainly reduces the load in the front and rear directions of the tower, and cannot effectively optimize the lateral load of the tower.

It can also reduce the load on the tower by suppressing the lateral deformation acceleration of the tower during the control process of the generator torque. However, the lateral deformation acceleration of the tower is mainly measured in real time through sensors, which has obvious nonlinear characteristics and is difficult to predict directly.

Or a multi-variable control strategy can be used to suppress the power output quality and the lateral oscillation of the tower to reduce the lateral load of the tower. However, the multi-objective cost function composed of nonlinear variables is difficult to solve, making the multi-variable control strategy difficult to implement.

Contents of the invention

In order to solve the above technical problems, the present disclosure provides a control method, system, electronic device and storage medium for a wind turbine.

The present disclosure solves the above technical problems through the following technical solutions:

In a first aspect, the present disclosure provides a control method for a wind turbine generator. The wind turbine includes a generator and a tower.

The steps of the control method include:

Obtain the inflow wind speed of the wind turbine generator in at least one predicted control period during the current control period;

The operating status data of the wind turbine in the predictive control period are obtained based on the incoming wind speed; wherein the operating status data includes predicted wind energy capture power, ideal wind energy capture power and lateral deformation of the tower acceleration;

Generate a target control sequence based on the operating status data and the preset cost function of the generator; wherein the target control sequence includes target torque values of the generator torque respectively corresponding to the predicted control period;

In response to entering the next control cycle, the generator torque is controlled according to the target torque value in the target control sequence.

Optionally, the preset cost function is expressed as:

in, Used to characterize the generator torque corresponding to the first predictive control period to the nth predictive control period, F ₁ is used to represent the first weight factor, and F ₂ is used to represent the second weight factor. , n is used to characterize the number of the predictive control cycles, T _s is used to characterize the duration of the predictive control cycle, _Pi is used to characterize the predicted wind energy capture power in the i-th predictive control cycle, /> Used to characterize the ideal wind energy capture power in the i-th predictive control period, /> Used to characterize the deformation acceleration in the i-th predictive control period, Used to characterize the preset maximum deformation acceleration record value;

The measured wind energy capture power, the ideal wind energy capture power and the deformation acceleration are all expressed based on the generator torque.

Optionally, the step of generating a target control sequence based on the operating status data and the preset cost function of the generator includes:

Use the particle swarm algorithm to iteratively calculate the torque control sequence and iterative change rate in multiple different solution directions, until the number of iterative calculations reaches the preset number of iterations, and obtain the target control sequence;

Wherein, the target control sequence is the torque control sequence corresponding to when the preset cost function takes a global minimum in all solution directions.

Optionally, the formula for the iterative calculation is:

in, Used to characterize the torque control sequence obtained by the jth iterative calculation,/> Used to characterize the iterative change rate obtained by the jth iterative calculation,/> Used to characterize the torque control sequence corresponding to when the preset cost function takes the minimum value in the first j iterations of the solution direction,/> Used to characterize the torque control sequence corresponding to when the preset cost function takes the global minimum in the first j iterations of all solution directions, c ₁ and c ₂ are used to characterize the learning factor, r ₁ and r ₂ are used to characterize the preset parameters respectively,/> Used to represent inertia weight.

Optionally, the wind turbine further includes a wind wheel;

The step of obtaining the operating status data of the wind turbine within the predictive control period based on the incoming wind speed includes:

According to the actual rotor speed, actual inflow wind speed, actual energy conversion efficiency and actual torque value in the previous control period of the current control period, as well as the inflow wind speed in each of the predicted control periods, The wind energy capture prediction model is used to obtain the rotor speed of the wind wheel within the predictive control period, as well as the predicted wind energy capture power and the ideal wind energy capture power of the wind turbine within the predictive control period.

Optionally, the step of obtaining the operating status data of the wind turbine within the predictive control period based on the incoming wind speed further includes:

According to the actual inflow wind speed, actual operating status data, and actual target torque value of the current control cycle and the previous control cycle, as well as the incoming wind speed and operating status data in each of the predicted control cycles, the deformation The acceleration prediction model sequentially predicts the deformation acceleration within the prediction control period.

Optionally, the wind energy capture prediction model is expressed as:

in, Used to characterize the predicted wind energy capture power in the i-th predicted control period, /> Used to characterize the ideal wind energy capture power in the i-th predictive control period, /> Used to characterize the energy conversion efficiency within the i-th predictive control cycle,/> The preset maximum value used to characterize the energy conversion efficiency,/> Used to characterize the incoming wind speed in the i-th predictive control period,/> is the theoretical value of the pitch angle corresponding to the incoming wind speed in the i-th predictive control period, and λ _i is used to characterize the tip speed ratio in the i-th predictive control period, /> is used to characterize the rotor speed of the wind wheel in the i-th predictive control period, N _gear is used to characterize the gearbox ratio of the wind turbine unit, J _rotor is used to characterize the rotational inertia of the wind wheel, ω ₀ is used to represent the actual rotor speed of the wind wheel in the current control period, and a ₁ , a ₂ , a ₃ , a ₄ , a ₅ , a ₆ , b ₁ , and b ₂ are used to represent the preset parameters respectively. .

Optionally, the deformation acceleration prediction model is expressed as:

Among them, F is used to characterize the deformation acceleration prediction model, is used to characterize the deformation acceleration in the i-th predictive control cycle, x _i is used to characterize the operating status data in the i-th predictive control cycle, Used to characterize the generator torque corresponding to the i-th predictive control cycle,/> It is used to characterize the incoming wind speed in the i-th predictive control period.

Optionally, the deformation acceleration prediction model is a support vector machine, and the kernel function in the deformation acceleration prediction model adopts a Gaussian kernel function; wherein the deformation acceleration prediction model is based on the historical operating status data and historical data of the wind turbine. It is trained with torque data and historical inflow wind speed data.

In a second aspect, the present disclosure provides a control system for a wind turbine generator. The wind turbine includes a generator and a tower.

The control system includes:

A wind speed acquisition module, configured to acquire the incoming wind speed of the wind turbine in at least one predicted control period within the current control period;

A state data prediction module, configured to obtain the operating state data of the wind turbine in the predictive control period based on the incoming wind speed; wherein the operating state data includes predicted wind energy capture power, ideal wind energy capture power and the Describe the lateral deformation acceleration of the tower;

A solution module, configured to generate a target control sequence based on the operating status data and the preset cost function of the generator; wherein the target control sequence includes generator torque corresponding to the predicted control period. Target torque value;

A control module configured to control the generator torque according to the target torque value in the target control sequence in response to entering the next control cycle.

Optionally, the preset cost function is expressed as:

in, Used to characterize the generator torque corresponding to the first predictive control period to the nth predictive control period, F ₁ is used to represent the first weight factor, and F ₂ is used to represent the second weight factor. , n is used to characterize the number of the predictive control cycles, T _s is used to characterize the duration of the predictive control cycle, _Pi is used to characterize the predicted wind energy capture power in the i-th predictive control cycle, /> Used to characterize the ideal wind energy capture power in the i-th predictive control period, /> Used to characterize the deformation acceleration in the i-th predictive control period, Used to characterize the preset maximum deformation acceleration record value;

The measured wind energy capture power, the ideal wind energy capture power and the deformation acceleration are all expressed based on the generator torque.

Optionally, the solution module is specifically configured to use the particle swarm algorithm to iteratively calculate the torque control sequence and the iterative change rate in multiple different solution directions, until the number of iterative calculations reaches a preset number of iterations, Obtain the target control sequence;

Wherein, the target control sequence is the torque control sequence corresponding to when the preset cost function takes a global minimum in all solution directions.

Optionally, the formula for the iterative calculation is:

in, Used to characterize the torque control sequence obtained by the jth iterative calculation,/> Used to characterize the iterative change rate obtained by the jth iterative calculation,/> Used to characterize the torque control sequence corresponding to when the preset cost function takes the minimum value in the first j iterations of the solution direction,/> Used to characterize the torque control sequence corresponding to when the preset cost function takes the global minimum in the first j iterations of all solution directions, c ₁ and c ₂ are used to characterize the learning factor, r ₁ and r ₂ are used to characterize the preset parameters respectively,/> Used to represent inertia weight.

Optionally, the wind turbine further includes a wind wheel;

The state data prediction module includes:

A wind energy capture prediction unit configured to calculate the actual rotor speed, the actual inflow wind speed, the actual energy conversion efficiency and the actual torque value in the previous control period of the current control period, and in each of the predicted control periods. of the incoming wind speed, using the wind energy capture prediction model to obtain the rotor speed of the wind wheel during the prediction control period, and the predicted wind energy capture power of the wind turbine during the prediction control period, the Ideal wind energy capture power.

Optionally, the state data prediction module includes:

The deformation acceleration prediction unit is used to calculate the actual inflow wind speed, actual operating status data, actual target torque value according to the current control period and the previous control period, and the inflow in each of the predicted control periods. Using the wind speed and operating status data, the deformation acceleration prediction model is used to predict the deformation acceleration within the prediction control period.

Optionally, the wind energy capture prediction model is expressed as:

in, Used to characterize the predicted wind energy capture power in the i-th predicted control period, /> Used to characterize the ideal wind energy capture power in the i-th predictive control period, /> Used to characterize the energy conversion efficiency within the i-th predictive control cycle,/> The preset maximum value used to characterize the energy conversion efficiency, Used to characterize the incoming wind speed in the i-th predictive control period,/> is the theoretical value of the pitch angle corresponding to the incoming wind speed in the i-th predictive control period, and λ _i is used to characterize the tip speed ratio in the i-th predictive control period, /> is used to characterize the rotor speed of the wind wheel in the i-th predictive control period, N _gear is used to characterize the gearbox ratio of the wind turbine unit, J _rotor is used to characterize the rotational inertia of the wind wheel, ω ₀ is used to represent the actual rotor speed of the wind wheel in the current control period, and a ₁ , a ₂ , a ₃ , a ₄ , a ₅ , a ₆ , b ₁ , and b ₂ are used to represent the preset parameters respectively. .

Optionally, the deformation acceleration prediction model is expressed as:

Among them, F is used to characterize the deformation acceleration prediction model, is used to characterize the deformation acceleration in the i-th predictive control cycle, x _i is used to characterize the operating status data in the i-th predictive control cycle, Used to characterize the generator torque corresponding to the i-th predictive control cycle,/> It is used to characterize the incoming wind speed in the i-th predictive control period.

Optionally, the deformation acceleration prediction model is a support vector machine, and the kernel function in the deformation acceleration prediction model adopts a Gaussian kernel function; wherein the deformation acceleration prediction model is based on the historical operating status data and historical data of the wind turbine. It is trained with torque data and historical inflow wind speed data.

In a third aspect, the present disclosure provides an electronic device, including a memory, a processor, and a computer program stored in the memory and used for running on the processor. When the processor executes the computer program, the control described in the first aspect is implemented. method.

In a fourth aspect, the present disclosure provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the control method described in the first aspect is implemented.

The positive progressive effect of this disclosure is that based on the Pareto theory, a preset cost function is proposed, which can collaboratively optimize the tower load of the wind turbine generator and the wind energy capture power, and can effectively reduce the lateral deflection of the tower of the wind turbine generator. Structural load, improving the power output quality of generators used in large wind turbines, and improving the reliability and energy efficiency of wind turbine operations have important engineering application value.

Moreover, combined with the established wind energy capture prediction model and the deformation acceleration prediction model used to obtain the lateral deformation acceleration of the tower, the particle swarm algorithm is used to solve the above-mentioned nonlinear preset cost function in real time to obtain the prediction control cycle including The target control sequence corresponding to the target torque value is used to control the generator torque in the next control cycle using the target torque value in the target control sequence.

Description of the drawings

Figure 1 is a schematic flow chart of a control method provided by Embodiment 1 of the present disclosure;

Figure 2 is a schematic flow chart of a particle swarm algorithm provided in Embodiment 1 of the present disclosure;

Figure 3 is a schematic module diagram of a control system provided in Embodiment 2 of the present disclosure;

Figure 4 is a schematic structural diagram of a wind turbine provided by Embodiment 3 of the present disclosure;

FIG. 5 is a schematic module diagram of an electronic device provided in Embodiment 4 of the present disclosure.

Detailed ways

The present disclosure is further described below by means of examples, but the present disclosure is not limited to the scope of the described examples.

It should be noted that if there are descriptions involving "first", "second", etc. in the embodiments of the present disclosure, the descriptions of "first", "second", etc. are only for descriptive purposes and cannot be understood as instructions or instructions. implying its relative importance or implicitly specifying the quantity of the technical feature indicated. Therefore, features defined as "first" and "second" may explicitly or implicitly include at least one of these features.

In addition, the technical solutions between the various embodiments can be combined with each other, but it must be based on the realization by those of ordinary skill in the art. When the combination of technical solutions is contradictory or cannot be realized, it should be considered that such a combination of technical solutions does not exist. , nor is it within the scope of protection required by this disclosure.

Example 1

This embodiment provides a control method for a wind turbine as shown in Figure 1 . Among them, wind turbines mainly include wind wheels (blades), generators and towers.

The control method includes the following steps:

S101. Obtain the inflow wind speed of the wind turbine generator in at least one predicted control period in the current control period;

S102. Obtain the operating status data of the wind turbine during the predictive control period based on the incoming wind speed; wherein the operating status data includes predicted wind energy capture power, ideal wind energy capture power and tower lateral deformation acceleration;

S103. Based on the operating status data and the preset cost function of the generator, generate a target control sequence; wherein the target control sequence includes target torque values of the generator torque corresponding to the predicted control period, and the number of target torque values corresponds to prediction step size;

S104. In response to entering the next control cycle, control the generator torque according to the target torque value in the target control sequence.

This embodiment predicts the operating status data of the wind turbine during the predictive control period by acquiring the inflow wind speed during the predictive control period. Among them, the operating status data includes the operating status data including predicted wind energy capture power, ideal wind energy capture power and tower lateral deformation acceleration.

And solve the preset cost function based on the operating status data to obtain the target control sequence. Wherein, the target control sequence includes each target torque value within the predictive control cycle.

After entering the next control cycle, the generator torque is controlled according to the target torque value in the target control sequence. This can achieve collaborative optimization of the wind turbine tower load and wind energy capture power, which can effectively reduce the lateral structural load of the wind turbine tower and improve the power output quality of the generator used in large wind turbines.

And after entering the next control period, use it as the current control period to repeat the above process to obtain the target control sequence again.

In step S101, there may be one or more predictive control periods, and the number of predictive control periods represents the prediction step size. That is, the prediction step size can control the length of the target control sequence. The number of target torque values contained in the target control sequence is the same as the number of prediction control cycles, and the target torque values correspond to the prediction control cycles one-to-one.

For example, first determine the prediction step size to be n, that is, determine the number of predictive control cycles to be n. The duration of each predictive control cycle is T _s . Taking into account the changes in wind speed and the accuracy of the wind measurement device, the prediction step size n and the length of the prediction control period T _s should not be too large.

Therefore, the inflow wind speed of the wind turbine from the 1st predictive control period to the nth predictive control period measured by the wind measuring device is:

The wind measuring device may be a laser radar detection device. LiDAR detection devices can calculate the distance of an object by sending a laser pulse and measuring its return time. In wind speed detection, the lidar detection device can measure the distance from the position of the wind turbine to particles in the air (such as dust, aerosols, etc.), and then determine the incoming wind speed that is about to reach the wind turbine.

The lidar detection device can obtain time series data of incoming wind speed through continuous measurements to represent the incoming wind speed measured continuously over a period of time. From this period of time, multiple discrete prediction control periods are determined, and the inflow wind speed in each prediction control period can be obtained respectively.

Based on step S101, the inflow wind speed from the 1st predictive control period to the nth predictive control period is obtained. Step S102 can separately obtain the operating status data of the wind turbine in each predictive control cycle.

Step S102 includes: utilizing the wind energy based on the actual rotor speed, actual inflow wind speed, actual energy conversion efficiency and actual torque value in the previous control period of the current control period, as well as the inflow wind speed in each predicted control period. The capture prediction model obtains the rotor speed of the wind wheel during the predictive control period, as well as the predicted wind energy capture power and ideal wind energy capture power of the wind turbine during the predictive control period.

For example, the wind energy capture prediction model can be expressed as:

in, Used to characterize the predicted wind energy capture power in the i-th predictive control cycle, P _i ^idea is used to characterize the ideal wind energy capture power in the i-th predictive control cycle,/> Used to characterize the energy conversion efficiency in the i-th predictive control cycle,/> Preset maximum value used to characterize energy conversion efficiency,/> Used to represent the inflow wind speed in the i-th predictive control period,/> is the theoretical value of the pitch angle corresponding to the incoming wind speed in the i-th predictive control cycle, λ _i is used to characterize the tip speed ratio in the i-th predictive control cycle,/> It is used to characterize the rotor speed of the wind wheel in the i-th predicted control period, N _gear is used to characterize the gearbox ratio of the wind turbine unit, J _rotor is used to characterize the rotational inertia of the wind wheel, and ω ₀ is used to characterize the current control period. The actual rotor speed of the inner wind wheel, a ₁ , a ₂ , a ₃ , a ₄ , a ₅ , a ₆ , b ₁ , and b ₂ are used to characterize the preset parameters respectively.

The theoretical value of the pitch angle β _i can be determined through the optimal wind-power curve. The optimal wind refers to the maximum wind energy capture power that can be achieved in the low wind speed area. This curve is determined during the design stage of the wind turbine.

For example, based on the actual rotor speed, actual inflow wind speed, actual energy conversion efficiency, and actual torque value of the wind turbine in the previous control cycle of the current control cycle, the wind turbine in the first predictive control cycle is calculated. Rotor speed, and wind energy conversion efficiency of the wind turbine in the first predictive control cycle;

Using the above wind energy capture prediction model, based on the incoming wind speed, the rotor speed in the first prediction control period and the wind energy conversion efficiency in the first prediction control period, the wind turbine's performance in each prediction control period is calculated sequentially. The rotor speed, the predicted wind energy capture power of the wind turbine in each predictive control cycle, and the ideal wind energy capture power of the wind turbine in each predictive control cycle.

That is to say, based on the inflow wind speed from the 1st predictive control period to the nth predictive control period With the above wind energy capture prediction model, the predicted wind energy capture power can be expressed as And the ideal wind energy capture power is expressed as/>

Step S102 also includes:

According to the actual inflow wind speed, actual operating status data, actual target torque value of the current control period and the previous control period, as well as the incoming wind speed and operating status data in each predictive control period, the deformation acceleration prediction model is used in sequence The deformation acceleration within the predictive control period is predicted.

Since the dynamic behavior of the wind turbine can be expressed as a second-order model, the second incoming wind speed of the predictive control period to be predicted can be obtained through at least two predicted predictive control periods.

For example, the deformation acceleration prediction model is expressed as:

Among them, F is used to characterize the deformation acceleration prediction model, is used to characterize the deformation acceleration in the i-th predictive control cycle, x _i is used to characterize the operating status data in the i-th predictive control cycle,/> Used to characterize the generator torque corresponding to the i-th predictive control cycle,/> Used to characterize the incoming wind speed in the i-th predictive control period.

Specifically, when predicting the deformation acceleration in the first predictive control period, the actual inflow wind speed, actual operating status data, actual target torque value, and the first predicted The first incoming wind speed of the control period is substituted into the above deformation acceleration prediction model, and the first deformation acceleration of the first prediction control period is predicted.

Similarly, when predicting the deformation acceleration in the second predictive control cycle, the first incoming wind speed, the first operating status data, the first target torque value of the first predictive control cycle, and the current control cycle can be used. The actual inflow wind speed, actual operating status data, and actual target torque value are substituted into the above-mentioned deformation acceleration prediction model, and the second deformation acceleration in the second prediction control cycle is predicted.

When predicting the deformation acceleration in the third and subsequent predictive control periods, the second incoming wind speed, second operating status data, second target torque value, and second predictive control period of the two first predictive control periods are combined. The third incoming wind speed of the period is substituted into the above deformation acceleration prediction model, and the third deformation acceleration of the second prediction control period is predicted.

The two first predictive control periods are two adjacent pre-stored control periods that have been predicted, and the second predictive control period is adjacent to the latter of the two first predictive control periods.

For example, when predicting the deformation acceleration in the first predictive control cycle:

For example, when predicting the deformation acceleration in the first predictive control cycle:

Among them, x _k0 and x _k-1 are used to represent the actual operating status data in the current control period and the previous control period respectively. are used to represent the actual torque value in the current control cycle and the previous control cycle, respectively. They are used to represent the actual inflow wind speed in the current control period and the previous control period respectively.

When predicting the deformation acceleration in the second predictive control cycle:

Moreover, the deformation acceleration prediction model may specifically be a support vector machine, and the kernel function in the deformation acceleration prediction model adopts a Gaussian kernel function.

Support vector regression is a machine learning algorithm used for pattern classification and regression analysis. It can effectively handle linear and nonlinear data classification problems.

The basic idea of SVR is to find an optimal hyperplane that can separate data samples of different categories, and to find a set of support vectors (support samples) on the boundaries on both sides of the hyperplane to define the decision boundary. This optimal hyperplane is called the maximum margin hyperplane, and its goal is to maximize the distance from the support vectors on the boundary to the decision boundary.

Among them, the deformation acceleration prediction model is trained based on the historical operating status data, historical torque data and historical incoming wind speed data of the wind turbine.

For example, the above deformation acceleration prediction model can be specifically expressed as:

Among them, F(y) is used to characterize the lateral deformation acceleration of the tower, W is used to characterize the multi-dimensional weight factor, and b is used to characterize the adjustable factor. It is used to characterize the regression equation to express the mapping relationship between input y and output F(y).

Substituting the Lagrangian multipliers ξ _i , ξ _i ^* and using the quadratic programming method to solve, the final tower lateral deformation acceleration prediction model can be expressed as:

Among them, K(y _i -y) is the Gaussian kernel function.

In step S103, the preset cost function can be expressed as:

The preset cost function is expressed as:

in, Used to characterize the generator torque corresponding to the 1st predictive control cycle to the nth predictive control cycle, F ₁ is used to represent the first weight factor, F ₂ is used to represent the second weight factor, and n is used to represent the prediction The number of control cycles, T _s is used to characterize the length of the predictive control cycle, Pi is used to represent the predicted wind energy capture power in the i-th predictive control cycle, and _{P i idea} ^is used to represent the ideal wind energy in the i-th predictive control cycle. capture power,/> Used to characterize the deformation acceleration in the i-th predictive control cycle,/> Used to characterize the preset maximum deformation acceleration record value.

The above measured wind energy capture power, ideal wind energy capture power and deformation acceleration are all expressed based on the generator torque.

In order to generate the target control sequence, the above-mentioned nonlinear preset cost function needs to be solved. Step S103 specifically includes:

According to the operating status data, the particle swarm algorithm is used to iteratively calculate the torque control sequence and iterative change rate in multiple different solution directions, until the number of iterative calculations reaches the preset number of iterations, and the target control sequence is obtained;

Among them, the target control sequence is the torque control sequence corresponding to when the preset cost function takes the global minimum in all solution directions.

For example, see Figure 2. Based on the particle swarm algorithm, the torque control sequence and the iterative change rate are iteratively calculated from m different solution directions, which is equivalent to searching in space for a group of m particles.

For any particle, various parameters used in the algorithm need to be initialized, such as learning factors and preset parameters. That is, for different solution directions, the values of the initialized learning factors and the preset parameters are different. Based on the initialized algorithm parameters, calculate the value of the preset cost function of each particle, update the iterative change rate and torque control sequence of each particle, and obtain the torque control corresponding to the minimum time based on the preset cost function calculated so far. sequence, updating the local optimal torque control sequence and the globally optimal torque control sequence. Then determine whether the number of iterative calculations is less than the preset number of iterations. If it is less, continue the iterative calculation to update the local optimal torque control sequence and the global optimal torque control sequence. Otherwise, output the global optimal torque control sequence. sequence as the target control sequence.

Then the preset cost function is iteratively calculated in different solution directions to obtain the value of the preset cost function.

The formula for iterative calculation is:

in, Used to characterize the torque control sequence obtained by the jth iterative calculation,/> Used to characterize the iterative change rate obtained by the jth iterative calculation,/> Used to characterize the torque control sequence corresponding to the minimum value of the preset cost function in the first j iterations of the solution direction,/> Used to characterize the corresponding torque control sequence when the preset cost function takes the global minimum in the first j iterations of all solution directions, c ₁ and c ₂ are used to characterize the learning factors respectively, r ₁ and r ₂ are used to represent the learning factors respectively. Characterize the preset parameters,/> Used to represent inertia weight.

Finally, the target control sequence is obtained

It should be noted that the iterative change rate includes change values corresponding to every n torque values in the corresponding torque control sequence. That is, the n torque values contained in the torque control sequence are calculated at the jth iteration, first obtain the iterative change rate calculated at the j+1 iteration, and compare each change value in the iterative change rate with the torque value. One-to-one corresponding addition can realize the j+1th iterative calculation of the torque control sequence.

In step S104, in response to entering the next control cycle, the first target torque value in the target control sequence is As the output of the generator controller, to control the generator torque.

In addition, before solving the above preset cost function, the target torque value of the generator torque of the wind turbine and the rotor speed of the wind turbine can also be constrained:

|ω ^ref -ω _i |≤σ _ω ω ^ref ,

in, It is used to characterize the torque reference value, ω ^ref is used to characterize the rotor speed reference value, σ _T is used to characterize the torque allowable deviation range, and σ _ω is used to characterize the speed allowable deviation range.

Example 2

This embodiment provides a wind turbine control system as shown in FIG. 3 . Among them, wind turbines mainly include wind wheels (blades), generators and towers.

The control system includes:

The wind speed acquisition module 301 is used to acquire the inflow wind speed of the wind turbine in at least one predicted control period in the current control period;

The state data prediction module 302 is used to obtain the operating state data of the wind turbine in the predictive control period based on the incoming wind speed; wherein the operating state data includes predicted wind energy capture power, ideal wind energy capture power and The lateral deformation acceleration of the tower;

The solution module 303 is configured to generate a target control sequence based on the operating status data and the preset cost function of the generator; wherein the target control sequence includes generator torques respectively corresponding to the predicted control periods. target torque value;

The control module 304 is configured to control the generator torque according to the torque value in the target control sequence in response to entering the next control cycle.

This embodiment predicts the operating status data of the wind turbine during the predictive control period by acquiring the inflow wind speed during the predictive control period. Among them, the operating status data includes the operating status data including predicted wind energy capture power, ideal wind energy capture power and tower lateral deformation acceleration.

And solve the preset cost function based on the operating status data to obtain the target control sequence. Wherein, the target control sequence includes each target torque value within the predictive control cycle.

After entering the next control cycle, the generator torque is controlled according to the target torque value in the target control sequence. This can achieve collaborative optimization of the wind turbine tower load and wind energy capture power, which can effectively reduce the lateral structural load of the wind turbine tower and improve the power output quality of the generator used in large wind turbines.

And after entering the next control period, use it as the current control period to repeat the above process to obtain the target control sequence again.

For the wind speed acquisition module 301, there may be one or more predictive control periods, and the number of predictive control periods represents the prediction step size. That is, the prediction step size can control the length of the target control sequence. The number of target torque values contained in the target control sequence is the same as the number of prediction control cycles, and the target torque values correspond to the prediction control cycles one-to-one.

For example, first determine the prediction step size to be n, that is, determine the number of predictive control cycles to be n. The duration of each predictive control cycle is T _s . Taking into account the changes in wind speed and the accuracy of the wind measurement device, the prediction step size n and the length of the prediction control period T _s should not be too large.

Therefore, the inflow wind speed of the wind turbine from the 1st predictive control period to the nth predictive control period measured by the wind measuring device is:

The wind measuring device may be a laser radar detection device. LiDAR detection devices can calculate the distance of an object by sending a laser pulse and measuring its return time. In wind speed detection, the lidar detection device can measure the distance from the position of the wind turbine to particles in the air (such as dust, aerosols, etc.), and then determine the incoming wind speed that is about to reach the wind turbine.

The lidar detection device can obtain time series data of incoming wind speed through continuous measurements to represent the incoming wind speed measured continuously over a period of time. From this period of time, multiple discrete prediction control periods are determined, and the inflow wind speed in each prediction control period can be obtained respectively.

Based on the wind speed acquisition module 301, the inflow wind speed from the 1st predictive control period to the nth predictive control period is obtained. The state data prediction module 302 can separately obtain the operating state data of the wind turbine generator in each predictive control cycle.

The state data prediction module 302 includes:

The wind energy capture prediction unit is used to calculate the actual rotor speed, actual inflow wind speed, actual energy conversion efficiency and actual torque value in the previous control period of the current control period, as well as the inflow wind speed in each predicted control period. , using the wind energy capture prediction model to obtain the rotor speed of the wind turbine during the predictive control period, as well as the predicted wind energy capture power and ideal wind energy capture power of the wind turbine unit during the predictive control period.

For example, the wind energy capture prediction model can be expressed as:

in, Used to characterize the predicted wind energy capture power in the i-th predictive control cycle, P _i ^idea is used to characterize the ideal wind energy capture power in the i-th predictive control cycle,/> Used to characterize the energy conversion efficiency in the i-th predictive control cycle,/> Preset maximum value used to characterize energy conversion efficiency,/> Used to represent the inflow wind speed in the i-th predictive control period,/> is the theoretical value of the pitch angle corresponding to the incoming wind speed in the i-th predictive control cycle, λ _i is used to characterize the tip speed ratio in the i-th predictive control cycle,/> It is used to characterize the rotor speed of the wind wheel in the i-th predicted control period, N _gear is used to characterize the gearbox ratio of the wind turbine unit, J _rotor is used to characterize the rotational inertia of the wind wheel, and ω ₀ is used to characterize the current control period. The actual rotor speed of the inner wind wheel, a ₁ , a ₂ , a ₃ , a ₄ , a ₅ , a ₆ , b ₁ , and b ₂ are used to characterize the preset parameters respectively.

The theoretical value of the pitch angle β _i can be determined through the optimal wind-power curve. The optimal wind refers to the maximum wind energy capture power that can be achieved in the low wind speed area. This curve is determined during the design stage of the wind turbine.

For example, based on the actual rotor speed, actual inflow wind speed, actual energy conversion efficiency, and actual torque value of the wind turbine in the previous control cycle of the current control cycle, the wind turbine in the first predictive control cycle is calculated. Rotor speed, and wind energy conversion efficiency of the wind turbine in the first predictive control cycle;

Using the above wind energy capture prediction model, based on the incoming wind speed, the rotor speed in the first prediction control period and the wind energy conversion efficiency in the first prediction control period, the wind turbine's performance in each prediction control period is calculated sequentially. The rotor speed, the predicted wind energy capture power of the wind turbine in each predictive control cycle, and the ideal wind energy capture power of the wind turbine in each predictive control cycle.

That is to say, based on the inflow wind speed from the 1st predictive control period to the nth predictive control period With the above wind energy capture prediction model, the predicted wind energy capture power can be expressed as And the ideal wind energy capture power is expressed as/>

The state data prediction module 302 also includes:

The deformation acceleration prediction unit is used to calculate the actual inflow wind speed, actual operating status data, and actual target torque value of the current control cycle and the previous control cycle, as well as the incoming wind speed and operating status data in each predicted control cycle. , using the deformation acceleration prediction model to sequentially predict the deformation acceleration within the prediction control period.

Since the dynamic behavior of the wind turbine can be expressed as a second-order model, the second incoming wind speed of the predictive control period to be predicted can be obtained through at least two predicted predictive control periods.

For example, the deformation acceleration prediction model is expressed as:

Among them, F is used to characterize the deformation acceleration prediction model, is used to characterize the deformation acceleration in the i-th predictive control cycle, x _i is used to characterize the operating status data in the i-th predictive control cycle,/> Used to characterize the generator torque corresponding to the i-th predictive control cycle,/> Used to characterize the incoming wind speed in the i-th predictive control period.

Specifically, when predicting the deformation acceleration in the first predictive control period, the actual inflow wind speed, actual operating status data, actual target torque value, and the first predicted The first incoming wind speed of the control period is substituted into the above deformation acceleration prediction model, and the first deformation acceleration of the first prediction control period is predicted.

Similarly, when predicting the deformation acceleration in the second predictive control cycle, the first incoming wind speed, the first operating status data, the first target torque value of the first predictive control cycle, and the current control cycle can be used. The actual inflow wind speed, actual operating status data, and actual target torque value are substituted into the above-mentioned deformation acceleration prediction model, and the second deformation acceleration in the second prediction control cycle is predicted.

When predicting the deformation acceleration in the third and subsequent predictive control periods, the second incoming wind speed, second operating status data, second target torque value, and second predictive control period of the two first predictive control periods are combined. The third incoming wind speed of the period is substituted into the above deformation acceleration prediction model, and the third deformation acceleration of the second prediction control period is predicted.

The two first predictive control periods are two adjacent pre-stored control periods that have been predicted, and the second predictive control period is adjacent to the latter of the two first predictive control periods.

For example, when predicting the deformation acceleration in the first predictive control period:

Among them, x _k0 and x _k-1 are used to represent the actual operating status data in the current control period and the previous control period respectively. are used to represent the actual torque value in the current control cycle and the previous control cycle, respectively. They are used to represent the actual inflow wind speed in the current control period and the previous control period respectively.

When predicting the deformation acceleration in the second predictive control cycle:

Moreover, the deformation acceleration prediction model may specifically be a support vector machine, and the kernel function in the deformation acceleration prediction model adopts a Gaussian kernel function.

Support Vector Regression (SVR) is a machine learning algorithm used for pattern classification and regression analysis. It can effectively handle linear and nonlinear data classification problems.

The basic idea of SVR is to find an optimal hyperplane that can separate data samples of different categories, and to find a set of support vectors (support samples) on the boundaries on both sides of the hyperplane to define the decision boundary. This optimal hyperplane is called the maximum margin hyperplane, and its goal is to maximize the distance from the support vectors on the boundary to the decision boundary.

Among them, the deformation acceleration prediction model is trained based on the historical operating status data, historical torque data and historical incoming wind speed data of the wind turbine.

For example, the above deformation acceleration prediction model can be specifically expressed as:

Among them, F(y) is used to characterize the lateral deformation acceleration of the tower, W is used to characterize the multi-dimensional weight factor, and b is used to characterize the adjustable factor. It is used to characterize the regression equation to express the mapping relationship between input y and output F(y).

Substituting the Lagrangian multipliers ξ _i , ξ _i ^* and using the quadratic programming method to solve, the final tower lateral deformation acceleration prediction model can be expressed as:

Among them, K(y _i -y) is the Gaussian kernel function.

The preset cost function is expressed as:

in, Used to characterize the generator torque corresponding to the 1st predictive control cycle to the nth predictive control cycle, F ₁ is used to represent the first weight factor, F ₂ is used to represent the second weight factor, and n is used to represent the prediction The number of control cycles, T _s is used to characterize the length of the predictive control cycle, Pi is used to represent the predicted wind energy capture power in the i-th predictive control cycle, and _{P i idea} ^is used to represent the ideal wind energy in the i-th predictive control cycle. capture power,/> Used to characterize the deformation acceleration in the i-th predictive control cycle,/> Used to characterize the preset maximum deformation acceleration record value.

The above measured wind energy capture power, ideal wind energy capture power and deformation acceleration are all expressed based on the generator torque.

In order to generate the target control sequence, the above-mentioned nonlinear preset cost function needs to be solved. The solution module 303 is specifically configured to use the particle swarm algorithm to iteratively calculate the torque control sequence and the iterative change rate in multiple different solution directions according to the operating status data, until the number of iterative calculations reaches the preset number of iterations, and the target is obtained. control sequence.

Among them, the target control sequence is the torque control sequence corresponding to when the preset cost function takes the global minimum in all solution directions.

For example, see Figure 2. Based on the particle swarm algorithm, the torque control sequence and the iterative change rate are iteratively calculated from m different solution directions, which is equivalent to searching in space for a group of m particles.

For any particle, various parameters used in the algorithm need to be initialized, such as learning factors and preset parameters. That is, for different solution directions, the values of the initialized learning factors and the preset parameters are different. Based on the initialized algorithm parameters, calculate the value of the preset cost function of each particle, update the iterative change rate and torque control sequence of each particle, and obtain the torque control corresponding to the minimum time based on the preset cost function calculated so far. sequence, updating the local optimal torque control sequence and the globally optimal torque control sequence. Then determine whether the number of iterative calculations is less than the preset number of iterations. If it is less, continue the iterative calculation to update the local optimal torque control sequence and the global optimal torque control sequence. Otherwise, output the global optimal torque control sequence. sequence as the target control sequence.

Then the preset cost function is iteratively calculated in different solution directions to obtain the value of the preset cost function.

The formula for iterative calculation is:

in, Used to characterize the torque control sequence obtained by the jth iterative calculation,/> Used to characterize the iterative change rate obtained by the jth iterative calculation,/> Used to characterize the torque control sequence corresponding to the minimum value of the preset cost function in the first j iterations of the solution direction,/> Used to characterize the corresponding torque control sequence when the preset cost function takes the global minimum in the first j iterations of all solution directions, c ₁ and c ₂ are used to characterize the learning factors respectively, r ₁ and r ₂ are used to represent the learning factors respectively. Characterize the preset parameters,/> Used to represent inertia weight.

Finally, the target control sequence is obtained

It should be noted that the iterative change rate includes change values corresponding to every n torque values in the corresponding torque control sequence. That is, the n torque values contained in the torque control sequence are calculated at the jth iteration, first obtain the iterative change rate calculated at the j+1 iteration, and compare each change value in the iterative change rate with the torque value. One-to-one corresponding addition can realize the j+1th iterative calculation of the torque control sequence.

In step S104, in response to entering the next control cycle, the first target torque value in the target control sequence is As the output of the generator controller, to control the generator torque.

In addition, before solving the above preset cost function, the target torque value of the generator torque of the wind turbine and the rotor speed of the wind turbine can also be constrained:

|ω ^ref -ω _i |≤σ _ω ω ^ref ,

in, It is used to characterize the torque reference value, ω ^ref is used to characterize the rotor speed reference value, σ _T is used to characterize the torque allowable deviation range, and σ _ω is used to characterize the speed allowable deviation range.

Example 3

This embodiment provides a wind turbine generator, including the control system in Embodiment 2.

See Figure 4 for an example. The incoming wind speed during the predictive control period obtained by the ultra-large wind turbine radar is used to predict the operating status data of the wind turbine during the predictive control period based on the incoming wind speed using the wind energy capture prediction model and the deformation acceleration prediction model. Among them, the operating status data includes the operating status data including predicted wind energy capture power, ideal wind energy capture power and tower lateral deformation acceleration. And use the particle swarm algorithm to solve the preset cost function based on the operating status data to obtain the target control sequence. Wherein, the target control sequence includes each target torque value within the predictive control cycle.

After entering the next control cycle, the generator torque is controlled according to the target torque value in the target control sequence. This can achieve collaborative optimization of the wind turbine tower load and wind energy capture power, which can effectively reduce the lateral structural load of the wind turbine tower and improve the power output quality of the generator used in large wind turbines. And after entering the next control period, use it as the current control period to repeat the above process to obtain the target control sequence again.

Example 4

FIG. 5 shows the structure of one electronic device of the present disclosure. The electronic device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the program, the above control method is implemented. The electronic device 50 shown in FIG. 5 is only an example and should not bring any limitations to the functions and scope of use of the embodiments of the present disclosure.

As shown in FIG. 5 , the electronic device 50 may also be in the form of a general computing device, for example, it may be a server device. The components of the electronic device 50 may include, but are not limited to: the above-mentioned at least one processor 51, the above-mentioned at least one memory 52, and a bus 53 connecting different system components (including the memory 52 and the processor 51).

Bus 53 includes a data bus, an address bus and a control bus.

Memory 52 may include volatile memory, such as random access memory (RAM) 521 and/or cache memory 522 , and may further include read-only memory (ROM) 523 .

Memory 52 may also include a program/utility 525 having a set of (at least one) program modules 524 including, but not limited to: an operating system, one or more application programs, other program modules, and program data. Each of the examples, or some combination thereof, may include the implementation of a network environment.

The processor 51 executes computer programs stored in the memory 52 to execute various functional applications and data processing, such as the above-mentioned control method of the present disclosure.

Electronic device 50 may also communicate with one or more external devices 54 (eg, keyboard, pointing device, etc.). This communication may occur through input/output (I/O) interface 55. Furthermore, the model generation device 50 may also communicate with one or more networks (eg, a local area network (LAN), a wide area network (WAN), and/or a public network, such as the Internet) through a network adapter 56 . As shown in FIG. 5 , network adapter 56 communicates with other modules of model-generated device 50 via bus 53 . It should be understood that, although not shown in the figures, other hardware and/or software modules may be used in conjunction with the model-generated device 50, including but not limited to: microcode, device drivers, redundant processors, external disk drive arrays, RAID (disk Array) systems, tape drives, and data backup storage systems, etc.

It should be noted that although several units/modules or sub-units/modules of the electronic device are mentioned in the above detailed description, this division is only exemplary and not mandatory. Indeed, according to embodiments of the present disclosure, the features and functions of two or more units/modules described above may be embodied in one unit/module. Conversely, the features and functions of one unit/module described above may be further divided to be embodied by multiple units/modules.

The present disclosure also provides a computer-readable storage medium on which a computer program is stored. When the program is executed by a processor, the above control method is implemented.

Among them, the readable storage medium that can be used may more specifically include but is not limited to: portable disk, hard disk, random access memory, read-only memory, erasable programmable read-only memory, optical storage device, magnetic storage device or any of the above. The right combination.

In a possible implementation, the present disclosure can also be implemented in the form of a program product, which includes program code. When the program product is run on a terminal device, the program code is used to cause the terminal device to execute the above control method.

The program code for executing the present disclosure can be written in any combination of one or more programming languages. The program code can be completely executed on the user device, partially executed on the user device, or as an independent software package. Executes, partially on the user device and partially on the remote device, or entirely on the remote device.

Although specific embodiments of the present disclosure have been described above, those skilled in the art will understand that these are only examples and the protection scope of the present disclosure is defined by the appended claims. Those skilled in the art can make various changes or modifications to these embodiments without departing from the principles and essence of the present disclosure, but these changes and modifications all fall within the protection scope of the present disclosure.

Claims (10)

1. The control method of the wind turbine generator is characterized in that the wind turbine generator comprises a generator and a tower;

the control method comprises the following steps:

acquiring the incoming wind speed of the wind turbine generator in at least one prediction control period in the current control period;

acquiring running state data of the wind turbine generator in the prediction control period respectively based on the incoming wind speed; wherein the operational state data includes predicted wind energy capture power, ideal wind energy capture power, and deformation acceleration of the tower side direction;

Generating a target control sequence based on the operating state data and a preset cost function of the generator; wherein the target control sequence includes target torque values of generator torque corresponding to the predicted control periods, respectively;

in response to entering a next control period, the generator torque is controlled in accordance with the target torque value in the target control sequence.

2. The control method according to claim 1, characterized in that the preset cost function is expressed as:

wherein ,for characterizing the generator torque respectively corresponding to the 1 st to nth predictive control periods, F ₁ For characterising the first weighting factor, F ₂ For characterizing a second weighting factor, n for characterizing the number of predictive control cycles, T _s For characterizing the duration of the predictive control period, P _i For characterizing said predicted wind energy capture power during the ith said predicted control period,/and/or->For characterizing said ideal wind energy capture power,/-during the ith said predictive control period>For characterizing said deformation acceleration in an ith said predictive control period,the method is used for representing a preset maximum deformation acceleration record value;

The measured wind energy capture power, the ideal wind energy capture power, and the deformation acceleration are all represented based on the generator torque.

3. The control method according to claim 2, wherein the step of generating a target control sequence based on the operating state data and a preset cost function of the generator comprises:

performing iterative computation on the torque control sequence and the iteration change rate in a plurality of different solving directions by using a particle swarm algorithm until the number of iterative computation reaches a preset iteration number, and acquiring the target control sequence;

the target control sequence is the torque control sequence corresponding to the preset cost function in all solving directions when the preset cost function takes a global minimum value.

4. A control method according to claim 3, wherein the formula of the iterative calculation is:

wherein ,for characterizing the torque control sequence obtained by the j-th iteration calculation,/and a method for controlling the torque control sequence>For characterizing said iterative rate of change obtained by the jth iterative calculation,/th iterative calculation>For characterizing said torque control sequence corresponding to when said preset cost function takes a minimum value in the first j iterative calculations of said solving direction,/- >C) representing the torque control sequence corresponding to the preset cost function taking the global minimum value in the previous j iterative calculations of all the solving directions ₁ and c₂ Respectively used for characterizing learning factors, r ₁ and r₂ Respectively used for representing preset parameters->For characterizing inertial weights.

5. A control method according to any one of claims 1-4, characterized in that the wind turbine further comprises a wind rotor;

the step of obtaining the running state data of the wind turbine generator set in the prediction control period based on the incoming wind speed comprises the following steps:

acquiring the rotor speed of the wind wheel in the prediction control period and the predicted wind energy capturing power and the ideal wind energy capturing power of the wind turbine in the prediction control period by using a wind energy capturing prediction model according to the actual rotor speed, the actual incoming wind speed, the actual energy conversion efficiency and the actual torque value in the control period which is the last control period of the current control period and the incoming wind speed in each prediction control period;

and/or the number of the groups of groups,

and according to the actual incoming flow wind speed, the actual running state data and the actual target torque value of the current control period and the previous control period, and the incoming flow wind speed and the running state data in each prediction control period, the deformation acceleration in the prediction control period is predicted by using a deformation acceleration prediction model in sequence.

6. The control method according to claim 5, wherein the wind energy capture prediction model is expressed as:

wherein ,for characterizing said predicted wind energy capture power during the ith said predicted control period,/and/or->For characterizing said ideal wind energy capture power,/-during the ith said predictive control period>For characterizing the energy conversion efficiency in the ith said predictive control cycle, +.>A preset maximum value for characterizing the energy conversion efficiency,/-for>For characterizing said incoming wind speed in the ith said predictive control period,/and/or->Lambda is the theoretical value of the pitch angle corresponding to the incoming wind speed in the ith predictive control period _i For characterizing the tip speed ratio in the ith said predictive control period,for characterising the rotor speed of the rotor during the ith said predictive control period, N _gear Gearbox transformation ratio for representing wind turbine generator system, J _rotor For characterising the moment of inertia, ω, of the rotor ₀ For characterizing the actual rotor speed of the rotor, a ₁ 、a ₂ 、a ₃ 、a ₄ 、a ₅ 、a ₆ 、b ₁ 、b ₂ Respectively used for representing preset parameters;

and/or the number of the groups of groups,

the deformation acceleration prediction model is expressed as:

wherein F is used for representing a deformation acceleration prediction model, For characterizing said deformation acceleration, x, in the ith said predictive control period _i For characterizing said operating state data in the ith said predictive control period,/and>for characterizing said generator torque corresponding to the ith said predictive control period,/and>for characterizing said incoming wind speed in an ith said predictive control period.

7. The control method according to claim 6, wherein the deformation acceleration prediction model is a support vector machine, and a gaussian kernel function is adopted as a kernel function in the deformation acceleration prediction model; the deformation acceleration prediction model is obtained through training according to historical running state data, historical torque data and historical incoming flow wind speed data of the wind turbine generator.

8. A control system of a wind turbine, wherein the wind turbine comprises a generator and a tower;

the control system includes:

the wind speed acquisition module is used for acquiring the incoming wind speed of the wind turbine generator in at least one prediction control period in the current control period;

the state data prediction module is used for acquiring running state data of the wind turbine generator in the prediction control period respectively based on the incoming wind speed; wherein the operational state data includes predicted wind energy capture power, ideal wind energy capture power, and deformation acceleration of the tower side direction;

The calculation module is used for generating a target control sequence based on the running state data and a preset cost function of the generator; wherein the target control sequence includes target torque values of generator torque corresponding to the predicted control periods, respectively;

and the control module is used for controlling the generator torque according to the target torque value in the target control sequence in response to entering the next control period.

9. An electronic device comprising a memory, a processor and a computer program stored on the memory for execution on the processor, characterized in that the processor implements the control method according to any one of claims 1-7 when executing the computer program.

10. A readable computer storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the control method according to any one of claims 1-7.