Control method and system of wind turbine generator, electronic device and storage medium

Abstract

The present disclosure provides a control method and system of a wind turbine generator, an electronic device and a storage medium. The control method comprises: obtaining the incoming wind speed of the wind turbine generator in at least one predicted control period in the current control period; obtaining the operating state data of the wind turbine generator in the predicted control period based on the incoming wind speed; generating a target control sequence based on the operating state data and a preset cost function of the generator; and controlling the generator torque according to the target torque value in the target control sequence in response to entering the next control period. The present disclosure proposes a preset cost function based on the Pareto theory, which can optimize the tower load and wind energy capture power of the wind turbine generator, effectively reduce the lateral structural load of the tower of the wind turbine generator, improve the power output quality of the generator applied in large-scale wind turbine generators, and improve the reliability and high energy efficiency of the wind turbine generator operation, which has important engineering application value.

Description

TECHNICAL FIELD

[0001] The present disclosure relates to the technical field of wind turbine control, in particular to a wind turbine control method and system, an electronic device and a storage medium. BACKGROUND

[0002] A wind turbine is a device that converts wind energy into electrical energy, which is composed of multiple components. It mainly includes a wind wheel (blade), a generator, a converter, a tower, and a control system. The wind wheel is driven to rotate by wind, activates the generator to generate electricity, and transmits the electricity to the power grid or other electrical equipment through the power transmission system.

[0003] With the increasing demand for wind energy, the rated power of the wind turbine and the size of the key supporting structure have been designed to be large, resulting in the obvious flexibility of the wind turbine structure.

[0004] Moreover, the influence of complex aerodynamic load of large wind turbines easily causes elastic deformation, thereby increasing the load of the tower and affecting the stability and service life of the wind turbine.

[0005] Currently, the structural damping of the tower can be adjusted by controlling the blade pitch angle and the generator torque. For example, in the control process of the blade pitch angle, an active damping control loop is added to reduce oscillation, thereby reducing the load of the tower without increasing additional structures. However, this method mainly reduces the load in the front and back directions of the tower, and cannot effectively optimize the load in the lateral direction of the tower.

[0006] In addition, in the control process of the generator torque, the load of the tower can be reduced by suppressing the lateral deformation acceleration of the tower. However, the lateral deformation acceleration of the tower is mainly measured in real time by a sensor, which has obvious nonlinear characteristics and is difficult to be directly predicted.

[0007] Alternatively, a multivariable control strategy can be used to suppress the power output quality and lateral oscillation of the tower to reduce the load in the lateral direction of the tower. However, the multi-objective cost function composed of nonlinear variables is difficult to solve, resulting in the difficulty of implementing the multivariable control strategy. SUMMARY

[0008] The present disclosure provides a wind turbine control method, system, electronic device and storage medium to solve the above technical problems.

[0009] The present disclosure solves the above technical problems by the following technical solutions:

[0010] In a first aspect, the present disclosure provides a wind turbine control method. The wind turbine includes a generator and a tower.

[0011] The steps of the control method include:

[0012] acquiring, in a current control period, a flow wind speed of the wind turbine in at least one predicted control period;

[0013] acquiring, based on the flow wind speed, operating state data of the wind turbine in the predicted control period respectively; wherein the operating state data comprises a predicted wind energy capture power, an ideal wind energy capture power and a deformation acceleration of the tower in lateral direction;

[0014] generating a target control sequence based on the operating state data and a preset cost function of the generator; wherein the target control sequence comprises target torque values of the generator torque corresponding to the predicted control period respectively;

[0015] in response to entering a next control period, controlling the generator torque according to the target torque values in the target control sequence.

[0016] Optionally, the preset cost function is expressed as:

[0017]

[0018] wherein, F1 is used to represent the generator torque corresponding to the first predicted control period, F2 is used to represent the second weight factor, n is used to represent the number of the predicted control period, T s is used to represent the time length of the predicted control period, is used to represent the predicted wind energy capture power in the i-th predicted control period, is used to represent the ideal wind energy capture power in the i-th predicted control period, is used to represent the deformation acceleration in the i-th predicted control period, is used to represent a preset maximum deformation acceleration record value;

[0019] The predicted wind energy capture power, the ideal wind energy capture power and the deformation acceleration are all represented based on the generator torque.

[0020] Optionally, the step of generating a target control sequence based on the operating state data and a preset cost function of the generator comprises:

[0021] using a particle swarm algorithm to perform iterative calculation on the torque control sequence and the iterative change rate in multiple different solving directions respectively until the number of the iterative calculation reaches a preset iteration number, to acquire the target control sequence;

[0022] The target control sequence is a torque control sequence corresponding to a global minimum of the preset cost function in all the solution directions.

[0023] Optionally, the iterative calculation formula is:

[0024]

[0025] wherein, is used to represent the torque control sequence obtained by the jth iteration calculation, is used to represent the iteration change rate obtained by the jth iteration calculation, is used to represent the torque control sequence corresponding to a minimum of the preset cost function in the first j iteration calculations of the solution direction, is used to represent the torque control sequence corresponding to a global minimum of the preset cost function in the first j iteration calculations of all the solution directions, c1 and c2 are respectively used to represent learning factors, and r1 and r2 are respectively used to represent preset parameters, is used to represent an inertia weight.

[0026] Optionally, the wind turbine further comprises a wind wheel;

[0027] The step of obtaining the operating state data of the wind turbine in the predicted control period based on the incoming flow wind speed comprises:

[0028] According to the actual rotor speed, the actual incoming flow wind speed, the actual energy conversion efficiency and the actual torque value in the last control period of the current control period, and the incoming flow wind speed in each predicted control period, a wind energy capture prediction model is used to obtain the rotor speed of the wind wheel in the predicted control period, and the predicted wind energy capture power and the ideal wind energy capture power of the wind turbine in the predicted control period.

[0029] Optionally, the step of obtaining the operating state data of the wind turbine in the predicted control period based on the incoming flow wind speed further comprises:

[0030] According to the actual incoming flow wind speed, the actual operating state data, the actual target torque value of the current control period and the last control period respectively, and the incoming flow wind speed and the operating state data in each predicted control period, a deformation acceleration prediction model is used to obtain the deformation acceleration in the predicted control period in sequence.

[0031] Optionally, the wind energy capture prediction model is represented as:

[0032]

[0033] wherein, ρi for representing the predicted wind energy capture power in the i-th prediction control period, a π for representing the air density, R for representing the radius of the wind turbine rotor, ρi for representing the ideal wind energy capture power in the i-th prediction control period, ηi for representing the energy conversion efficiency in the i-th prediction control period, ηmax for representing the preset maximum value of the energy conversion efficiency, Vi for representing the inflow wind speed in the i-th prediction control period, θi for representing the theoretical value of the pitch angle corresponding to the inflow wind speed in the i-th prediction control period, λi for representing the tip speed ratio value in the i-th prediction control period, Ni for representing the rotor speed of the wind turbine in the i-th prediction control period, gear J for representing the gear box ratio of the wind turbine, rotor I for representing the moment of inertia of the wind turbine, ω0 for representing the actual rotor speed of the wind turbine in the current control period, a1, a2, a3, a4, a5, a6, b1, b2 for representing preset parameters, respectively.

[0034] Optionally, the deformation acceleration prediction model is represented as:

[0035]

[0036]

[0037] wherein, F represents the deformation acceleration prediction model, Fi for representing the deformation acceleration in the i-th prediction control period, xi for representing the operating state data in the i-th prediction control period, i xi for representing the actual operating state data in the i-th current control period, Ti for representing the generator torque corresponding to the i-th prediction control period, Vi for representing the inflow wind speed in the i-th prediction control period, i Pi for representing the active power of the wind turbine corresponding to the i-th prediction control period, i ωi for representing the rotor speed of the wind turbine corresponding to the i-th prediction control period.

[0038] Optionally, the deformation acceleration prediction model is a support vector machine, and a kernel function in the deformation acceleration prediction model adopts a Gaussian kernel function; wherein the deformation acceleration prediction model is trained according to historical operation state data, historical torque data and historical incoming flow wind speed data of the wind turbine.

[0039] In a second aspect, the present disclosure provides a control system of a wind turbine. The wind turbine includes a generator and a tower.

[0040] The control system includes:

[0041] A wind speed acquisition module is configured to acquire incoming flow wind speed of the wind turbine in at least one predicted control period in a current control period;

[0042] A state data prediction module is configured to acquire operation state data of the wind turbine in the predicted control period based on the incoming flow wind speed; wherein the operation state data includes predicted wind energy capture power, ideal wind energy capture power and deformation acceleration of the tower side;

[0043] A solving module is configured to generate a target control sequence based on the operation state data and a preset cost function of the generator; wherein the target control sequence includes target torque values of the generator torque corresponding to the predicted control period respectively;

[0044] A control module is configured to control the generator torque according to the target torque values in the target control sequence in response to entering a next control period.

[0045] Optionally, the preset cost function is expressed as:

[0046]

[0047] wherein, F1 is used to represent the generator torque corresponding to the first predicted control period, F2 is used to represent the second weight factor, n is used to represent the number of the predicted control period, T s is used to represent the time length of the predicted control period, is used to represent the predicted wind energy capture power in the i th predicted control period, is used to represent the ideal wind energy capture power in the i th predicted control period, is used to represent the deformation acceleration in the i th predicted control period, is used to represent a preset maximum deformation acceleration record value;

[0048] The predicted wind energy capture power, the ideal wind energy capture power, and the deformation acceleration are represented based on the generator torque.

[0049] Optionally, the solving module is specifically configured to use a particle swarm algorithm to iteratively calculate the torque control sequence and the iteration change rate in multiple different solving directions, until a preset iteration number of the iterative calculation is reached, to obtain the target control sequence.

[0050] Optionally, the target control sequence is the torque control sequence corresponding to a global minimum value of the preset cost function in all the solving directions.

[0051] Optionally, a formula of the iterative calculation is as follows:

[0052]

[0053] wherein, is used to represent the torque control sequence obtained by the jth iteration calculation, is used to represent the iteration change rate obtained by the jth iteration calculation, is used to represent the torque control sequence corresponding to a minimum value of the preset cost function in the first j iteration calculations in the solving direction, is used to represent the torque control sequence corresponding to a global minimum value of the preset cost function in the first j iteration calculations in all the solving directions, c1 and c2 are respectively used to represent learning factors, and r1 and r2 are respectively used to represent preset parameters, is used to represent an inertia weight.

[0054] Optionally, the wind turbine further includes a wind wheel.

[0055] The state data prediction module includes:

[0056] The wind energy capture prediction unit is configured to obtain rotor rotation speed of the wind wheel in the prediction control period and the predicted wind energy capture power and the ideal wind energy capture power of the wind turbine in the prediction control period by using a wind energy capture prediction model according to actual rotor rotation speed, actual incoming flow wind speed, actual energy conversion efficiency, and actual torque value in a previous control period of the current control period and the incoming flow wind speed in each prediction control period.

[0057] Optionally, the state data prediction module includes:

[0058] a deformation acceleration prediction unit configured to predict a deformation acceleration in each of the prediction control periods according to the actual inflow wind speed, the actual operating state data, the actual target torque value in each of the current control period and the previous control period, and the inflow wind speed and the operating state data in each of the prediction control periods, by using a deformation acceleration prediction model.

[0059] Optionally, the wind energy capture prediction model is represented as:

[0060]

[0061] wherein, ρi is used to represent the predicted wind energy capture power in the i-th prediction control period, a π is used to represent the air density, and R is used to represent the radius of the wind wheel of the wind turbine, ηi is used to represent the ideal wind energy capture power in the i-th prediction control period, ηi is used to represent the energy conversion efficiency in the i-th prediction control period, ηmax is used to represent a preset maximum value of the energy conversion efficiency, Vi is used to represent the inflow wind speed in the i-th prediction control period, θi is used to represent a theoretical value of the pitch angle corresponding to the inflow wind speed in the i-th prediction control period, λi is used to represent a tip speed ratio value in the i-th prediction control period, Ni is used to represent the rotor speed of the wind wheel in the i-th prediction control period, gear J is used to represent the gear box ratio of the wind turbine, rotor I is used to represent the rotational inertia of the wind wheel, ω0 is used to represent the actual rotor speed of the wind wheel in the current control period, and a1, a2, a3, a4, a5, a6, b1, b2 are used to represent preset parameters, respectively.

[0062] Optionally, the deformation acceleration prediction model is represented as:

[0063]

[0064] wherein, F is used to represent the deformation acceleration prediction model, Fi is used to represent the deformation acceleration in the i-th prediction control period, xi is used to represent the operating state data in the i-th prediction control period, i xi is used to represent the actual operating state data in the i-th current control period, Ti is used to represent the generator torque corresponding to the i-th prediction control period, P is used to represent the wind speed of the i th prediction control cycle i ω is used to represent the active power of the wind turbine corresponding to the i th prediction control cycle i ω is used to represent the wind wheel speed of the wind turbine corresponding to the i th prediction control cycle.

[0065] Optionally, the deformation acceleration prediction model is a support vector machine, and a kernel function in the deformation acceleration prediction model adopts a Gaussian kernel function; wherein the deformation acceleration prediction model is trained according to historical running state data, historical torque data and historical wind speed data of the wind turbine.

[0066] In a third aspect, the present disclosure provides an electronic device, comprising a memory, a processor and a computer program stored in the memory and used to run on the processor, and when the processor executes the computer program, the control method of the first aspect is realized.

[0067] In a fourth aspect, the present disclosure provides a computer readable storage medium, which stores a computer program, and when the processor executes the computer program, the control method of the first aspect is realized.

[0068] The positive progress effect of the present disclosure is that based on the Pareto theory, a preset cost function is proposed, which can optimize the tower load of the wind turbine and the wind energy capture power, can effectively reduce the lateral structural load of the tower of the wind turbine, improve the power output quality of the generator applied in large-scale wind turbines, and improve the reliability and high energy efficiency of the wind turbine operation, which has important engineering application value.

[0069] Moreover, combined with the established wind energy capture prediction model and the deformation acceleration prediction model for obtaining the deformation acceleration of the tower side, a particle swarm algorithm is adopted to solve the above-mentioned nonlinear preset cost function in real time to obtain a target control sequence including a target torque value corresponding to each prediction control cycle, and then the target torque value in the target control sequence is used to control the generator torque in the next control cycle. BRIEF DESCRIPTION OF DRAWINGS

[0070] Figure 1 A flowchart of a control method provided for Embodiment 1 of the present disclosure is shown in the figure;

[0071] Figure 2 A flowchart of a particle swarm algorithm provided for Embodiment 1 of the present disclosure is shown in the figure;

[0072] Figure 3 A module diagram of a control system provided for Embodiment 2 of the present disclosure is shown in the figure;

[0073] Figure 4 A structural schematic diagram of a wind turbine provided for Embodiment 3 of the present disclosure;

[0074] Figure 5 A module schematic diagram of an electronic device provided for Embodiment 4 of the present disclosure. DETAILED DESCRIPTION

[0075] The present disclosure is further illustrated by the following examples, but the present disclosure is not limited in scope within the examples described.

[0076] It should be noted that if the present disclosure has a description involving "first", "second", etc. in the embodiments, the description of "first", "second", etc. is only for the purpose of description, and cannot be understood as indicating or implying the relative importance of the indicated technical features or implicitly indicating the number of the indicated technical features. Therefore, the features limited by "first", "second" can be explicitly or implicitly included at least one of the features.

[0077] In addition, the technical solutions of each embodiment can be combined with each other, but it must be based on the realization of the ordinary skilled in the art, when the combination of technical solutions appears contradictory or unachievable, it should be considered that the combination of technical solutions does not exist, nor within the protection scope required by the present disclosure.

[0078] Embodiment 1

[0079] The present embodiment provides a control method of a wind turbine as shown in Figure 1 The wind turbine mainly includes a wind wheel (blade), a generator and a tower, etc.

[0080] The control method comprises the following steps:

[0081] S101, obtaining the incoming flow wind speed of the wind turbine in at least one prediction control period in the current control period;

[0082] S102, obtaining the running state data of the wind turbine in the prediction control period respectively based on the incoming flow wind speed; wherein the running state data includes the predicted wind energy capture power, the ideal wind energy capture power and the deformation acceleration of the tower side;

[0083] S103, generating a target control sequence based on the running state data and the preset cost function of the generator; wherein the target control sequence includes the target torque value of the generator torque corresponding to the prediction control period respectively, and the number of target torque values corresponds to the prediction step;

[0084] S104, in response to entering the next control period, controlling the generator torque according to the target torque value in the target control sequence.

[0085] The embodiment predicts the operation state data of the wind turbine in the prediction control period by the obtained incoming flow wind speed in the prediction control period. The operation state data includes the predicted wind energy capture power, the ideal wind energy capture power, and the deformation acceleration of the tower in the lateral direction.

[0086] The preset cost function is solved based on the operation state data to obtain a target control sequence. The target control sequence includes a target torque value in each prediction control period.

[0087] In the next control period, the generator torque is controlled according to the target torque value in the target control sequence. The tower load and the wind energy capture power of the wind turbine are cooperatively optimized, which can effectively reduce the lateral structural load of the tower of the wind turbine and improve the power output quality of the generator applied in large-scale wind turbines.

[0088] The above process is repeated as the current control period to obtain the target control sequence again after entering the next control period.

[0089] In step S101, there can be one or more prediction control periods, and the number of prediction control periods represents the prediction step. That is, the prediction step can control the length of the target control sequence, and the number of target torque values contained in the target control sequence is the same as the number of prediction control periods, and the target torque value corresponds to the prediction control period one by one.

[0090] For example, the prediction step is determined to be n, that is, the number of prediction control periods is determined to be n. The length of each prediction control period is T s . Considering the wind speed change and the accuracy of the wind measuring device, the prediction step n and the length of the prediction control period T s should not be too large.

[0091] Therefore, the incoming flow wind speed of the wind turbine from the first prediction control period to the nth prediction control period is measured by the wind measuring device in turn as

[0092] The wind measuring device can be a laser radar detection device. The laser radar detection device can calculate the distance of an object by sending a laser pulse and measuring the return time. In wind speed detection, the laser radar detection device can measure the distance from the location of the wind turbine to the particulate matter (such as dust, aerosol, etc.) in the air, and then determine the incoming flow wind speed approaching the wind turbine.

[0093] The laser radar detection device can obtain time series data of the incoming flow wind speed by continuous measurement to represent the incoming flow wind speed measured continuously in a period of time. A plurality of discrete prediction control periods are determined from the period of time, and the incoming flow wind speed in each prediction control period is obtained.

[0094] Based on the step S101, the incoming wind speed from the first prediction control period to the nth prediction control period is obtained The step S102 can obtain the operating state data of the wind turbine in each prediction control period respectively.

[0095] The step S102 comprises: obtaining the rotor speed of the wind wheel in the prediction control period, and the predicted wind energy capture power and the ideal wind energy capture power of the wind turbine in the prediction control period, by using the wind energy capture 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 last control period of the current control period, and the incoming wind speed in each prediction control period.

[0096] For example, the wind energy capture prediction model can be expressed as:

[0097]

[0098]

[0099] wherein, ρi is used to represent the predicted wind energy capture power in the ith prediction control period, a π is used to represent the circular constant, R is used to represent the radius of the wind wheel of the wind turbine, βi is used to represent the ideal wind energy capture power in the ith prediction control period, ηi is used to represent the energy conversion efficiency in the ith prediction control period, ηmax is used to represent the preset maximum value of the energy conversion efficiency, Vi is used to represent the incoming wind speed in the ith prediction control period, βi is the theoretical value of the pitch angle corresponding to the incoming wind speed in the ith prediction control period, λi is used to represent the tip speed ratio value in the ith prediction control period, Ni is used to represent the rotor speed of the wind wheel in the ith prediction control period, gear J is used to represent the gear box ratio of the wind turbine, rotor I is used to represent the rotational inertia of the wind wheel, ω0 is used to represent the actual rotor speed of the wind wheel in the current control period, and a1, a2, a3, a4, a5, a6, b1, b2 are used to represent preset parameters respectively.

[0100] The theoretical value of the pitch angle β i The optimal wind can be determined by the optimal wind-power curve, which is determined in the design stage of the wind turbine, and the optimal wind refers to the maximum wind energy capture power that can be achieved in the low wind speed area.

[0101] For example, the rotor speed of the wind wheel in the first predicted control period and the wind energy conversion efficiency of the wind turbine in the first predicted control period are calculated according to the actual rotor speed, the actual incoming wind speed, the actual energy conversion efficiency, and the actual torque value of the wind wheel in the last control period of the current control period.

[0102] By using the wind energy capture prediction model, the rotor speed of the wind wheel in each predicted control period, the predicted wind energy capture power of the wind turbine in each predicted control period, and the ideal wind energy capture power of the wind turbine in each predicted control period are sequentially calculated based on the incoming wind speed, the rotor speed in the first predicted control period, and the wind energy conversion efficiency in the first predicted control period.

[0103] That is, the incoming wind speed in the first predicted control period to the nth predicted control period and the wind energy capture prediction model can be expressed as and the ideal wind energy capture power is respectively expressed as

[0104] Step S102 further includes:

[0105] According to the actual incoming wind speed, the actual operating state data, the actual target torque value of the current control period and the last control period, and the incoming wind speed and operating state data in each predicted control period, the deformation acceleration in the predicted control period is sequentially predicted by using the deformation acceleration prediction model.

[0106] Since the dynamic behavior of the wind turbine can be represented as a second-order model, the second incoming wind speed of the predicted control period to be predicted can be obtained by at least two predicted predicted control periods.

[0107] For example, the deformation acceleration prediction model is expressed as:

[0108]

[0109] wherein F is used to represent the deformation acceleration prediction model, is used to represent the deformation acceleration in the ith predicted control period, is used to represent the operating state data in the ith predicted control period, x i is used to represent the actual operating state data in the ith current control period, is used to represent the generator torque corresponding to the ith predicted control period, is used to represent the incoming wind speed in the ith predicted control period, P iω i ω

[0110] Specifically, when predicting the deformation acceleration of the first prediction control period, the actual incoming wind speed, the actual operating state data, the actual target torque value of the current control period and the last control period, and the first incoming wind speed of the first prediction control period are substituted into the above deformation acceleration prediction model to predict the first deformation acceleration of the first prediction control period.

[0111] Similarly, when predicting the deformation acceleration of the second prediction control period, the first incoming wind speed, the first operating state data, the first target torque value of the first prediction control period, and the actual incoming wind speed, the actual operating state data, the actual target torque value of the current control period are substituted into the above deformation acceleration prediction model to predict the second deformation acceleration of the second prediction control period.

[0112] When predicting the deformation acceleration of the third and subsequent prediction control periods, the second incoming wind speed, the second operating state data, the second target torque value of the two first prediction control periods, and the third incoming wind speed of the second prediction control period are substituted into the above deformation acceleration prediction model to predict the third deformation acceleration of the second prediction control period.

[0113] Wherein, the two first prediction control periods are two adjacent pre-existing control periods that have been predicted, and the second prediction control period is adjacent to the latter of the two first prediction control periods.

[0114] For example, when predicting the deformation acceleration of the first prediction control period:

[0115] For example, when predicting the deformation acceleration of the first prediction control period:

[0116]

[0117] Wherein, x k0 , x k-1 respectively represent the actual operating state data in the current control period and the last control period, respectively represent the actual torque value in the current control period and the last control period, respectively represent the actual incoming wind speed in the current control period and the last control period.

[0118] When predicting the deformation acceleration of the second prediction control period:

[0119]

[0120] And the deformation acceleration prediction model can be a support vector machine, and the kernel function in the deformation acceleration prediction model can be a Gaussian kernel function.

[0121] Support vector regression is a machine learning algorithm for pattern classification and regression analysis. It can effectively handle linear and nonlinear data classification problems.

[0122] The basic idea of SVR is to find an optimal hyperplane that can separate different classes of data samples, and find a set of support vectors (support samples) on the boundary of the hyperplane to define the decision boundary. This optimal hyperplane is called the maximum margin hyperplane, which aims to maximize the distance of support vectors on the boundary to the decision boundary.

[0123] The deformation acceleration prediction model is trained according to historical operating state data, historical torque data and historical incoming flow wind speed data of the wind turbine.

[0124] For example, the deformation acceleration prediction model can be specifically represented as:

[0125]

[0126] Where F(y) is used to represent the deformation acceleration of the tower side, W is used to represent the multi-dimensional weight factor, and b is used to represent the adjustable factor, to represent the regression equation to express the mapping relationship between input y and output F(y).

[0127] The Lagrange multiplier ξ i ,ξ i * is substituted and solved by using the quadratic programming method, and finally the deformation acceleration prediction model of the tower side can be represented as:

[0128]

[0129] Where K(y i -y) is a Gaussian kernel function.

[0130] In step S103, the preset cost function can be represented as:

[0131] The preset cost function is represented as:

[0132]

[0133] Where, F1 is used to represent a first weight factor, F2 is used to represent a second weight factor, n is used to represent a number of prediction control periods, T s is used to represent a time length of a prediction control period, is used to represent a predicted wind energy capture power in the i-th prediction control period, is used to represent an ideal wind energy capture power in the i-th prediction control period, is used to represent a deformation acceleration in the i-th prediction control period, is used to represent a preset maximum deformation acceleration record value.

[0134] The predicted wind energy capture power, the ideal wind energy capture power and the deformation acceleration are all represented based on the generator torque.

[0135] In order to generate the target control sequence, the above-mentioned nonlinear preset cost function needs to be solved. Step S103 specifically comprises:

[0136] According to the operating state data, the particle swarm algorithm is used to iteratively calculate the torque control sequence and the iterative change rate in multiple different solving directions until the number of iterative calculations reaches a preset iteration number, so as to obtain the target control sequence.

[0137] The target control sequence is the torque control sequence corresponding to the global minimum value of the preset cost function in all solving directions.

[0138] For example, referring to Figure 2 , based on the particle swarm algorithm, the torque control sequence and the iterative change rate are iteratively calculated from m different solving directions, which is equivalent to searching in space by a group consisting of m particles.

[0139] For any particle, various parameters used in the algorithm need to be initialized, such as learning factors and preset parameters. That is, for different solving directions, the values of the learning factors and the preset parameters obtained by initialization are different. Based on the initialized algorithm parameters, the value of the preset cost function of each particle is calculated, the iterative change rate and the torque control sequence of each particle are updated, the torque control sequence corresponding to the minimum value of the preset cost function currently calculated is updated, the locally optimal torque control sequence and the globally optimal torque control sequence are updated. Then, it is judged whether the number of iterative calculations is less than the preset iteration number. If it is less, the iterative calculation is continued to update the locally optimal torque control sequence and the globally optimal torque control sequence, otherwise, the globally optimal torque control sequence is output as the target control sequence.

[0140] Then, the preset cost function is iteratively calculated in different solving directions to obtain a value of the preset cost function.

[0141] The formula of the iterative calculation is:

[0142]

[0143] wherein, is used to represent a torque control sequence obtained by the jthiterative calculation, is used to represent an iterative change rate obtained by the jthiterative calculation, is used to represent a torque control sequence corresponding to a minimum value of the preset cost function in the first j iterative calculations in the solving direction, is used to represent a torque control sequence corresponding to a global minimum value of the preset cost function in the first j iterative calculations in all solving directions, and c1and c2are respectively used to represent learning factors, and r1and r2are respectively used to represent preset parameters, is used to represent an inertia weight.

[0144] Finally, a target control sequence

[0145] It should be noted that the iterative change rate includes a change value corresponding to each of n torque values in the corresponding torque control sequence. That is, the n torque values contained in the torque control sequence obtained by the jthiterative calculation are added to each change value in the iterative change rate obtained by the j+1thiterative calculation one by one, so that the j+1thiterative calculation of the torque control sequence is realized.

[0146] In step S104, in response to entering a next control period, a first target torque value in the target control sequence is output as a controller of the generator to realize control on the generator torque.

[0147] In addition, before solving the above preset cost function, the target torque value of the generator torque of the wind turbine generator and the rotor speed of the wind wheel can also be constrained:

[0148]

[0149] |ω ref -ω i |≤σ ω ω ref ,

[0150] wherein, is used to represent a torque reference value, ω ref is used to represent a rotor speed reference value, σ T is used to represent a torque allowable deviation range, σω for characterizing the rotational speed allowable deviation range.

[0151] Embodiment 2

[0152] The embodiment provides a control system of a wind turbine as shown in the drawings. Figure 3 The wind turbine mainly comprises a wind wheel (blade), a generator and a tower, etc.

[0153] The control system comprises:

[0154] A wind speed acquisition module 301 is configured to acquire, in a current control period, a flow wind speed of the wind turbine in at least one predicted control period;

[0155] A state data prediction module 302 is configured to acquire, based on the flow wind speed, running state data of the wind turbine in the predicted control period respectively; wherein the running state data comprises a predicted wind energy capture power, an ideal wind energy capture power and a deformation acceleration of a lateral side of the tower;

[0156] A solving module 303 is configured to generate a target control sequence based on the running state data and a preset cost function of the generator; wherein the target control sequence comprises a target torque value of a generator torque corresponding to the predicted control period respectively;

[0157] A control module 304 is configured to control the generator torque according to the torque value in the target control sequence in response to entering a next control period.

[0158] The embodiment predicts the running state data of the wind turbine in the predicted control period by the acquired flow wind speed in the predicted control period. The running state data comprises a predicted wind energy capture power, an ideal wind energy capture power and a deformation acceleration of a lateral side of the tower.

[0159] And the preset cost function is solved based on the running state data to obtain the target control sequence. The target control sequence comprises a target torque value in each predicted control period.

[0160] After entering the next control period, the generator torque is controlled according to the target torque value in the target control sequence. The tower load and the wind energy capture power of the wind turbine are cooperatively optimized, the lateral structural load of the tower of the wind turbine is effectively reduced, and the power output quality of the generator applied in the large wind turbine is improved.

[0161] And after entering the next control period, the above process is repeated as the current control period to obtain the target control sequence again.

[0162] For the wind speed acquisition module 301, there can be one or more prediction control periods, and the number of prediction control periods represents the prediction step. That is, the prediction step can control the length of the target control sequence, and the target control sequence contains the same number of target torque values as the number of prediction control periods, and the target torque value corresponds to the prediction control period one by one.

[0163] For example, first determine the prediction step n, that is, determine the number of prediction control periods n. The length of each prediction control period is T s . Considering the wind speed change and the accuracy of the wind measuring device, the prediction step n and the length of the prediction control period T s should not be too large.

[0164] Therefore, the incoming flow wind speed of the wind turbine from the first prediction control period to the nth prediction control period measured by the wind measuring device is

[0165] Wherein, the wind measuring device can be a laser radar detection device. The laser radar detection device can calculate the distance of an object by sending a laser pulse and measuring its return time. In wind speed detection, the laser radar detection device can measure the distance from the location of the wind turbine to the particulate matter (such as dust, aerosol, etc.) in the air, and then determine the incoming flow wind speed that will reach the wind turbine.

[0166] The laser radar detection device can obtain time series data of the incoming flow wind speed by continuous measurement to represent the incoming flow wind speed measured continuously in a period of time. A plurality of discrete prediction control periods are determined from this period of time, that is, the incoming flow wind speed in each prediction control period can be obtained.

[0167] Based on the wind speed acquisition module 301, the incoming flow wind speed from the first prediction control period to the nth prediction control period is The state data prediction module 302 can acquire the operating state data of the wind turbine in each prediction control period.

[0168] The state data prediction module 302 includes:

[0169] The wind energy capture prediction unit is configured to acquire the rotor speed of the wind wheel in the prediction control period, and the predicted wind energy capture power and the ideal wind energy capture power of the wind turbine in the prediction control period, by using a wind energy capture prediction model according to the actual rotor speed, the actual incoming flow wind speed, the actual energy conversion efficiency and the actual torque value in the last control period of the current control period, and the incoming flow wind speed in each prediction control period.

[0170] For example, the wind energy capture prediction model can be represented as:

[0171]

[0172] wherein, ρi is used to represent the predicted wind energy capture power in the i-th prediction control period, a π is used to represent the circular constant, R is used to represent the radius of the wind wheel of the wind turbine, ρi is used to represent the predicted wind energy capture power in the i-th prediction control period, ηi is used to represent the energy conversion efficiency in the i-th prediction control period, ηmax is used to represent the preset maximum value of the energy conversion efficiency, Vi is used to represent the incoming wind speed in the i-th prediction control period, βi is used to represent the theoretical value of the pitch angle corresponding to the incoming wind speed in the i-th prediction control period, λi is used to represent the tip speed ratio value in the i-th prediction control period, Ni is used to represent the rotor speed of the wind wheel in the i-th prediction control period, gear J is used to represent the gear box ratio of the wind turbine, rotor I is used to represent the moment of inertia of the wind wheel, ω0 is used to represent the actual rotor speed of the wind wheel in the current control period, a1, a2, a3, a4, a5, a6, b1, b2 are respectively used to represent preset parameters.

[0173] the theoretical value of the pitch angle β i The optimal wind refers to the maximum wind energy capture power that can be achieved in the low wind speed region, and the curve is determined in the design stage of the wind turbine.

[0174] For example, based on the actual rotor speed of the wind wheel in the last control period of the current control period, the actual incoming wind speed, the actual energy conversion efficiency, and the actual torque value, the rotor speed of the wind wheel in the first prediction control period and the wind energy conversion efficiency of the wind turbine in the first prediction control period are calculated.

[0175] 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 rotor speed of the wind wheel in each prediction control period, the predicted wind energy capture power of the wind turbine in each prediction control period, and the ideal wind energy capture power of the wind turbine in each prediction control period are sequentially calculated.

[0176] That is, based on the incoming wind speed from the first prediction control period to the n-th prediction control period and the above wind energy capture prediction model, the predicted wind energy capture power can be represented as and the ideal wind energy capture power can be represented as

[0177] The state data prediction module 302 further comprises:

[0178] a deformation acceleration prediction unit, configured to sequentially predict the deformation acceleration in each prediction control period by using a deformation acceleration prediction model according to the actual inflow wind speed, the actual operating state data, the actual target torque value of each of the current control period and the previous control period, and the inflow wind speed and the operating state data in each prediction control period.

[0179] Since the dynamic behavior of the wind turbine can be represented as a second-order model, the second inflow wind speed of the prediction control period to be predicted can be obtained through at least two predicted prediction control periods.

[0180] For example, the deformation acceleration prediction model is represented as:

[0181]

[0182] wherein F is used to represent the deformation acceleration prediction model, is used to represent the deformation acceleration in the i th prediction control period, is used to represent the operating state data in the i th prediction control period, x i is used to represent the actual operating state data in the i th current control period, is used to represent the generator torque corresponding to the i th prediction control period, is used to represent the inflow wind speed in the i th prediction control period, P i is used to represent the active power of the wind turbine corresponding to the i th prediction control period, ω i is used to represent the rotor speed of the wind turbine corresponding to the i th prediction control period.

[0183] Specifically, when predicting the deformation acceleration of the first prediction control period, the actual inflow wind speed, the actual operating state data, the actual target torque value of each of the current control period and the previous control period, and the first inflow wind speed of the first prediction control period are substituted into the above deformation acceleration prediction model to predict the first deformation acceleration of the first prediction control period.

[0184] Similarly, when predicting the deformation acceleration of the second prediction control period, the first inflow wind speed, the first operating state data, and the first target torque value of the first prediction control period, and the actual inflow wind speed, the actual operating state data, and the actual target torque value of the current control period are substituted into the above deformation acceleration prediction model to predict the second deformation acceleration of the second prediction control period.

[0185] When predicting the deformation acceleration of the third prediction control period, the second incoming flow wind speed, the second operating state data, the second target torque value of the two first prediction control periods, and the third incoming flow wind speed of the second prediction control period are substituted into the deformation acceleration prediction model to predict the third deformation acceleration of the second prediction control period.

[0186] The two first prediction control periods are two adjacent pre-stored control periods that have been predicted, and the second prediction control period is adjacent to the latter of the two first prediction control periods.

[0187] For example, when predicting the deformation acceleration of the first prediction control period:

[0188]

[0189] wherein x k0 , x k-1 respectively represent the actual operating state data in the current control period and the previous control period, respectively represent the actual torque value in the current control period and the previous control period, respectively represent the actual incoming flow wind speed in the current control period and the previous control period.

[0190] When predicting the deformation acceleration of the second prediction control period:

[0191]

[0192] Furthermore, the deformation acceleration prediction model can be a support vector machine, and the kernel function in the deformation acceleration prediction model can be a Gaussian kernel function.

[0193] 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.

[0194] The basic idea of SVR is to find an optimal hyperplane that can separate different classes of data samples, and find a set of support vectors (support samples) on the boundary 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 of support vectors on the boundary to the decision boundary.

[0195] The deformation acceleration prediction model is trained according to historical operating state data, historical torque data, and historical incoming flow wind speed data of the wind turbine.

[0196] For example, the deformation acceleration prediction model can be specifically represented as:

[0197]

[0198] wherein F(y) is used to represent the deformation acceleration of the tower in lateral direction, W is used to represent the multi-dimensional weight factor, and b is used to represent the adjustable factor, is used to represent the regression equation to express the mapping relationship between the input y and the output F(y).

[0199] The Lagrange multiplier ξ i ,ξ i * is substituted, and a quadratic programming method is used to solve, and finally the deformation acceleration prediction model of the tower in lateral direction can be expressed as:

[0200]

[0201] wherein K(y i -y) is a Gaussian kernel function.

[0202] The preset cost function is expressed as:

[0203]

[0204] wherein, is used to represent the generator torque corresponding to the first prediction control period to the nth prediction control period respectively, F1 is used to represent the first weight factor, F2 is used to represent the second weight factor, n is used to represent the number of prediction control periods, T s is used to represent the length of the prediction control period, P i is used to represent the predicted wind energy capture power in the i-th prediction control period, is used to represent the ideal wind energy capture power in the i-th prediction control period, is used to represent the deformation acceleration in the i-th prediction control period, is used to represent the preset maximum deformation acceleration record value.

[0205] The above-mentioned wind energy capture power, ideal wind energy capture power and deformation acceleration are all expressed based on the generator torque.

[0206] In order to generate the target control sequence, the above-mentioned nonlinear preset cost function needs to be solved. The solving module 303 is specifically configured to use a particle swarm algorithm to iteratively calculate the torque control sequence and the iterative change rate in multiple different solving directions according to the operating state data, until the number of iterative calculations reaches a preset iteration number, and the target control sequence is obtained.

[0207] wherein the target control sequence is the torque control sequence corresponding to the global minimum value of the preset cost function in all solving directions.

[0208] For example, see Figure 2 Based on the particle swarm algorithm, the torque control sequence and the iteration change rate are iteratively calculated from m different solving directions, which is equivalent to searching in space by a group consisting of m particles.

[0209] For any particle, various parameters used in the algorithm need to be initialized, such as the learning factor and the preset parameter. That is, for different solving directions, the values of the learning factor and the preset parameter obtained by initialization are different. Based on the initialized algorithm parameters, the value of the preset cost function of each particle is calculated, the iteration change rate and the torque control sequence of each particle are updated, the torque control sequence corresponding to the minimum preset cost function is updated based on the currently calculated preset cost function, the locally optimal torque control sequence and the globally optimal torque control sequence are updated. Then, it is judged whether the number of iteration calculations is less than the preset iteration number. If it is less, the iteration calculation is continued to update the locally optimal torque control sequence and the globally optimal torque control sequence, otherwise, the globally optimal torque control sequence is output as the target control sequence.

[0210] Then, the preset cost function is iteratively calculated in different solving directions to obtain the value of the preset cost function.

[0211] The formula for iteration calculation is:

[0212]

[0213]

[0214] wherein, is used to represent the torque control sequence obtained by the jth iteration calculation, is used to represent the iteration change rate obtained by the jth iteration calculation, is used to represent the torque control sequence corresponding to the minimum preset cost function in the first j iteration calculations in the solving direction, is used to represent the torque control sequence corresponding to the global minimum preset cost function in the first j iteration calculations in all solving directions, c1 and c2 are used to represent the learning factor, and r1 and r2 are used to represent the preset parameter, is used to represent the inertia weight.

[0215] Finally, the target control sequence

[0216] It should be noted that the iteration change rate includes a change value corresponding to each n torque value in the corresponding torque control sequence. That is, n torque values contained in the torque control sequence are calculated by the jth iteration, each change value in the iteration change rate is added one by one corresponding to the torque value, and the j+1th iteration calculation of the torque control sequence is realized.

[0217] In step S104, in response to entering the next control period, the first target torque value in the target control sequence is output as the output of the controller of the generator, so as to realize control on the generator torque. As the output of the controller of the generator, so as to realize control on the generator torque.

[0218] In addition, before solving the above-mentioned preset cost function, the target torque value of the generator torque of the wind turbine generator and the rotor speed of the wind wheel can also be constrained:

[0219]

[0220] |ω ref -ω i |≤σ ω ω ref ,

[0221] Wherein, is used to represent the torque reference value, ω ref is used to represent the rotor speed reference value, σ T is used to represent the torque allowable deviation range, σ ω is used to represent the speed allowable deviation range.

[0222] Embodiment 3

[0223] The embodiment provides a wind turbine generator comprising the control system in embodiment 2.

[0224] For example, Figure 4 The inflow wind speed in the prediction control period obtained by the radar of the super large wind turbine generator is used to predict the running state data of the wind turbine generator in the prediction control period by using the wind energy capture prediction model and the deformation acceleration prediction model according to the inflow wind speed. The running state data includes the predicted wind energy capture power, the ideal wind energy capture power and the deformation acceleration of the tower side. And the particle swarm algorithm is used to solve the preset cost function based on the running state data to obtain the target control sequence. The target control sequence includes the target torque value in each prediction control period.

[0225] After entering the next control period, the generator torque is controlled according to the target torque value in the target control sequence. That is, the tower load of the wind turbine and the wind energy capture power are cooperatively optimized, the lateral structural load of the tower of the wind turbine is effectively reduced, and the power output quality of the generator applied in the large-scale wind turbine is improved. After entering the next control period, the above process is repeated as the current control period to obtain the target control sequence again.

[0226] Embodiment 4

[0227] Figure 5 The structure of one kind of electronic device in the present disclosure is shown. The electronic device comprises a memory, a processor, and a computer program stored in the memory and executable on the processor, and the processor implements the above control method when executing the program. Figure 5 The electronic device 50 shown is only an example and should not limit the function and use range of the embodiments of the present disclosure.

[0228] As Figure 5 shown, the electronic device 50 can also be in the form of a general computing device, for example, it can be a server device. The components of the electronic device 50 can 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.

[0229] The bus 53 includes a data bus, an address bus, and a control bus.

[0230] The memory 52 can include volatile memory, such as a random access memory (RAM) 521 and / or a cache memory 522, and can further include a read-only memory (ROM) 523.

[0231] The memory 52 can also include programs / utilities 525 having a set of (at least one) program modules 524, such as an operating system, one or more application programs, other program modules, and program data, each of which or some combination of which can include the implementation of a network environment.

[0232] The processor 51 performs various function applications and data processing by running the computer program stored in the memory 52, such as the above-mentioned control method of the present disclosure.

[0233] The electronic device 50 can also communicate with one or more external devices 54 such as a keyboard or a pointing device, by way of Input / Output (I / O) interfaces 55. Further, the model generation device 50 can communicate to one or more networks such as a local area network (LAN), a wide area network (WAN), and / or the public network, such as the Internet, by way of the network adapter 56. As Figure 5 illustrated, the network adapter 56 communicates to the other components of the model generation device 50 by way of the bus 53. It is to be appreciated that a more

[0234] It is to be noted that although several means / modules or sub-means / modules of the electronic device are mentioned in the foregoing detailed description, such division into means / modules or sub-means / modules is merely exemplary and not mandatory. Indeed, according to an embodiment of the disclosure, the features and functionalities of two or more of the above-described means / modules can be embodied in a single means / module. Conversely, a single one of the above-described means / modules can be split into several means / modules each embodying a subset of the features and functionalities of the single means / module.

[0235] The disclosure also provides a computer readable storage medium having stored thereon a computer program, the program being executable by a processor to implement the above control method.

[0236] More particularly, the computer readable storage medium can include, but is not limited to, portable discs, hard disks, random access memories, read-only memories, erasable programmable read-only memories, optical storage devices, magnetic storage devices, or any suitable combination of the foregoing.

[0237] In a possible implementation, the disclosure can also be implemented in the form of a program product, which includes program codes for causing an end device to execute the above control method when the program product is run on the end device.

[0238] The program codes for implementing the disclosure can be written in any combination of one or more programming languages, and can be entirely executed on the user device, partially executed on the user device, executed as a stand-alone software package, partially executed on the user device and partially on a remote device, or entirely executed on a remote device.

[0239] Although the specific embodiments of the present disclosure are described above, it should be understood by those skilled in the art that this is only an example, 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 the embodiments without departing from the principles and essence of the present disclosure, and such changes and modifications all fall within the protection scope of the present disclosure.

Claims

1. A method of controlling a wind turbine, characterized by, The wind turbine comprises a generator and a tower; The steps of the control method comprise: obtaining the incoming flow wind speed of the wind turbine in at least one predicted control period in the current control period; obtaining the operating state data of the wind turbine in the predicted control period respectively based on the incoming flow wind speed; wherein the operating state data comprises predicted wind energy capture power, ideal wind energy capture power and deformation acceleration of the tower side; generating a target control sequence based on the operating state data and a preset cost function of the generator; wherein the target control sequence comprises target torque values of the generator torque corresponding to the predicted control period respectively; in response to entering the next control period, controlling the generator torque according to the target torque values in the target control sequence; The preset cost function is represented as: wherein, F1 is used to represent a first weight factor, F2 is used to represent a second weight factor, n is used to represent a number of the prediction control periods, T s is used to represent a time length of the prediction control period, is used to represent the predicted wind energy capture power in the i-th prediction control period, is used to represent the ideal wind energy capture power in the i-th prediction control period, is used to represent the deformation acceleration in the i-th prediction control period, is used to represent a preset maximum deformation acceleration record value; The predicted wind energy capture power, the ideal wind energy capture power and the deformation acceleration are all represented based on the generator torque.

2. The control method according to claim 1, characterized by, The step of generating a target control sequence based on the operating state data and a preset cost function of the generator comprises: using a particle swarm algorithm to iteratively calculate the torque control sequence and the iteration rate in multiple different solution directions until the number of iterations reaches a preset iteration number, and obtaining the target control sequence; Wherein, the target control sequence is the torque control sequence corresponding to the global minimum value of the preset cost function in all solution directions.

3. The control method according to claim 2, characterized by, The formula of the iterative calculation is: wherein, for representing the torque control sequence obtained by the jth iteration calculation, for representing the iteration rate obtained by the jth iteration calculation, for representing the torque control sequence corresponding to the minimum value of the preset cost function in the first j iteration calculations of the solving direction, for representing the torque control sequence corresponding to the global minimum value of the preset cost function in the first j iteration calculations of all the solving directions, and c1 and c2 are respectively used for representing learning factors, and r1 and r2 are respectively used for representing preset parameters, for representing an inertia weight.

4. The control method according to any one of claims 1 to 3, characterized by, The wind turbine further comprises a wind wheel; The step of obtaining the operating state data of the wind turbine in the predicted control period based on the incoming flow wind speed comprises: According to the actual rotor speed, actual incoming flow wind speed, actual energy conversion efficiency and actual torque value in the last control period of the current control period, and the incoming flow wind speed in each predicted control period, a wind energy capture prediction model is used to obtain the rotor speed of the wind wheel in the predicted control period, and the predicted wind energy capture power, the ideal wind energy capture power of the wind turbine in the predicted control period; And / or, According to the actual incoming flow wind speed, actual operating state data and actual target torque value of the current control period and the last control period respectively, and the incoming flow wind speed and operating state data in each predicted control period, a deformation acceleration prediction model is used to sequentially predict the deformation acceleration in the predicted control period.

5. The control method according to claim 4, characterized by The wind energy capture prediction model is represented as: wherein, for representing the predicted wind energy capture power in the i-th prediction control period, a for representing the air density, p for representing the circular constant, R for representing the rotor radius of the wind turbine, for representing the ideal wind energy capture power in the i-th prediction control period, for representing the energy conversion efficiency in the i-th prediction control period, for representing the preset maximum value of the energy conversion efficiency, for representing the inflow wind speed in the i-th prediction control period, for representing the theoretical value of the pitch angle corresponding to the inflow wind speed in the i-th prediction control period, for representing the tip speed ratio in the i-th prediction control period, for representing the rotor speed of the wind wheel in the i-th prediction control period, N gear for representing the gear box ratio of the wind turbine, J rotor for representing the moment of inertia of the wind wheel, ω0 for representing the actual rotor speed of the wind wheel in the current control period, a1, a2, a3, a4, a5, a6, b1, b2 for representing preset parameters, respectively. And / or, The deformation acceleration prediction model is represented as: wherein F is used to represent a deformation acceleration prediction model, is used to represent the deformation acceleration in the i-th prediction control period, is used to represent the operating state data in the i-th prediction control period, i is used to represent the actual operating state data in the i-th current control period, is used to represent the generator torque corresponding to the i-th prediction control period, is used to represent the incoming flow wind speed in the i-th prediction control period, i is used to represent the active power of the wind turbine corresponding to the i-th prediction control period, i is used to represent the rotor speed of the wind turbine corresponding to the i-th prediction control period.

6. The control method according to claim 5, characterized by 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 trained according to the historical operating state data, historical torque data and historical incoming flow wind speed data of the wind turbine.

7. A control system for a wind turbine generator characterized by, The wind turbine comprises a generator and a tower; The control system comprises: a wind speed acquisition module, configured to obtain the incoming flow wind speed of the wind turbine in at least one predicted control period in the current control period; a state data prediction module, configured to acquire operation state data of the wind turbine in the prediction control period based on the incoming flow wind speed, wherein the operation state data comprises predicted wind energy capture power, ideal wind energy capture power, and deformation acceleration of the tower in lateral direction; a solving module, configured to generate a target control sequence based on the operation state data and a preset cost function of the generator, wherein the target control sequence comprises target torque values of the generator torque corresponding to the prediction control period respectively; a control module, configured to control the generator torque according to the target torque values in the target control sequence in response to entering a next control period; the preset cost function is expressed as: wherein, F1 is used to represent a first weight factor, F2 is used to represent a second weight factor, n is used to represent a number of the prediction control periods, T s is used to represent a time length of the prediction control period, is used to represent the predicted wind energy capture power in the i-th prediction control period, is used to represent the ideal wind energy capture power in the i-th prediction control period, is used to represent the deformation acceleration in the i-th prediction control period, is used to represent a preset maximum deformation acceleration record value; the predicted wind energy capture power, the ideal wind energy capture power, and the deformation acceleration are all expressed based on the generator torque.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory for running on the processor, characterized in that, the processor implements the control method as claimed in any one of claims 1-6 when executing the computer program.

9. A computer readable storage medium having stored thereon a computer program, characterized in that, the computer program implements the control method as claimed in any one of claims 1-6 when executed by the processor.

Images