Offshore wind turbine generator yaw control method, device and equipment based on MPC and medium

Abstract

The invention provides an offshore wind turbine generator yaw control method, device and equipment based on MPC and a medium. The control method comprises the steps of performing multi-time scale average value processing and variational mode decomposition on acquired historical wind speed and wind direction data, training wind speed and wind direction training data based on a convolutional neural network, a long-short term memory network and a self-attention mechanism layer, and constructing a wind speed and wind direction prediction model; predicting the average wind speed and wind direction of the next time period according to the current wind speed and wind direction and the prediction model; training the predicted average wind speed and wind direction and the current cabin angle, and constructing an MPC prediction control model; the current angular velocity and angle are substituted into a prediction control model for prediction and optimization, and a target yaw angular velocity and a target yaw angle are obtained; the angular velocity is adjusted to be the target yaw angular velocity, and the orientation angle is adjusted to be the target yaw angle; the method can accurately predict the wind speed and the wind direction, quickly respond to the wind direction change and improve the yaw control precision.

Description

Technical Field

[0001] The present invention relates to the technical field of yaw control, and in particular to an MPC-based offshore wind turbine yaw control method, device, equipment and medium. Background Art

[0002] In the field of wind power generation, the yaw control method is the core technology to ensure the efficient and stable operation of wind turbines. The wind speed and direction information of the wind field is captured by the wind direction sensor, and the wind speed and direction information data is transmitted to the yaw controller. After receiving the data, the yaw controller comprehensively considers key factors such as the operating status of the wind turbine and accurately calculates the angle at which the wind turbine needs to turn. Subsequently, the yaw controller issues a steering command to the wind direction driver, and the wind direction driver adjusts the direction of the wind turbine body and wind rotor according to the command to ensure that the blades of the wind turbine are always aligned with the wind direction.

[0003] Traditional yaw control methods are mainly divided into yaw control methods based on traditional wind direction sensors and yaw control methods based on wind direction prediction. The yaw control method based on traditional wind direction sensors usually relies on real-time wind direction sensor data, but these data may be affected by factors such as environmental noise, sensor accuracy limitations, and data transmission delays, resulting in low accuracy of wind direction data; while the yaw control method based on wind direction prediction adjusts the direction of the wind turbine in advance by predicting future wind direction changes. However, since this method predicts future wind direction changes based on historical data, this prediction may have errors. When there is a deviation between the actual wind direction and the predicted result, the traditional yaw control method based on wind direction prediction may not be able to make adjustments in time, resulting in a deviation between the direction of the wind turbine and the actual wind direction. Summary of the invention

[0004] The present invention aims to solve at least one of the technical problems existing in the prior art. To this end, the present invention proposes a method, device, equipment and medium for yaw control of an offshore wind turbine based on MPC, which can accurately predict wind speed and wind direction, quickly respond to wind direction changes, and improve the accuracy of yaw control.

[0005] In a first aspect, an embodiment of the present invention provides an offshore wind turbine yaw control method based on MPC, comprising:

[0006] Obtain multiple sets of historical wind speed data and multiple sets of historical wind direction data;

[0007] The multiple sets of historical wind speed data and the multiple sets of historical wind direction data are respectively subjected to multi-time scale average processing and then subjected to variational mode decomposition to obtain multiple sets of wind speed training data and multiple sets of wind direction training data of different frequency items corresponding to different time scales;

[0008] Based on a convolutional neural network, a long short-term memory network and a self-attention mechanism layer, model training and construction are performed on the multiple groups of wind speed training data and the multiple groups of wind direction training data of different frequency items corresponding to different time scales to obtain a wind speed and direction prediction model;

[0009] According to the current wind speed and the wind speed and wind direction prediction model, the average wind speed of the next period is predicted; according to the current wind direction and the wind speed and wind direction prediction model, the average wind direction angle of the next period is predicted;

[0010] Model training and construction are performed according to the average wind speed of the next period, the average wind direction angle of the next period, and the orientation angle of the offshore wind turbine cabin at the current moment to obtain an MPC predictive control model;

[0011] Based on a preset time step, the current cabin rotation angular velocity and the current cabin heading angle are substituted into the MPC predictive control model for prediction and optimization to obtain a target yaw angular velocity and a target yaw angle;

[0012] According to the target yaw angular velocity and the target yaw angle, the angular velocity of the offshore wind turbine generator set is adjusted to the target yaw angular velocity, and the orientation angle of the offshore wind turbine generator set is adjusted to the target yaw angle.

[0013] In some embodiments of the present invention, the multiple sets of historical wind speed data and the multiple sets of historical wind direction data are respectively subjected to multi-time scale average processing and then subjected to variational mode decomposition to obtain multiple sets of wind speed training data and multiple sets of wind direction training data with different frequency terms corresponding to different time scales, including:

[0014] Performing multi-time scale average processing on the multiple sets of historical wind speed data and the multiple sets of historical wind direction data, respectively, to obtain multiple sets of average wind speed data and multiple sets of average wind direction data corresponding to different time scales;

[0015] The multiple groups of average wind speed data and the multiple groups of average wind direction data corresponding to different time scales are calculated based on the variational mode decomposition formula to obtain wind speed frequency item data and wind direction frequency item data corresponding to different time scales;

[0016] The wind speed frequency item data and the wind direction frequency item data corresponding to different time scales are processed respectively by a sliding window rolling method to obtain multiple groups of wind speed training data and multiple groups of wind direction training data of different frequency items corresponding to different time scales.

[0017] In some embodiments of the present invention, the multiple sets of historical wind speed data and the multiple sets of historical wind direction data are respectively subjected to multi-time scale average processing to obtain multiple sets of average wind speed data and multiple sets of average wind direction data corresponding to different time scales, including:

[0018] Dividing the multiple groups of historical wind speed data and the multiple groups of historical wind direction data according to different time scales, respectively, to obtain multiple groups of first wind speed data and multiple groups of first wind direction data corresponding to different time scales;

[0019] Averaging the multiple groups of first wind speed data corresponding to different time scales to obtain multiple groups of average wind speed data corresponding to different time scales;

[0020] The multiple groups of first wind direction data corresponding to different time scales are averaged to obtain multiple groups of average wind direction data corresponding to different time scales.

[0021] In some embodiments of the present invention, the wind speed frequency item data includes wind speed high frequency item data, wind speed medium frequency item data, wind speed low frequency item data and wind speed residual item data; the wind direction frequency item data includes wind direction high frequency item data, wind direction medium frequency item data, wind direction low frequency item data and wind direction residual item data;

[0022] The variational mode decomposition formula is:

[0023]

[0024] Among them, the value of k is a positive integer, which represents the number of types of frequency data; u k is the intrinsic mode function of each modal function, namely, the wind speed high frequency item data, wind speed medium frequency item data, wind speed low frequency item data and wind direction high frequency item data, wind direction medium frequency item data, wind direction low frequency item data; ω k is the center frequency of each intrinsic mode function; δ(t) is the Dirac function; f is the average wind speed data or the average wind direction data; the wind speed residual term data is the residual frequency data after subtracting the extracted wind speed high-frequency term data, wind speed medium-frequency term data, and wind speed low-frequency term data from the average wind speed data; the wind direction residual term data is the residual frequency data after subtracting the extracted wind direction high-frequency term data, wind direction medium-frequency term data, and wind direction low-frequency term data from the average wind direction data.

[0025] In some embodiments of the present invention, the wind speed frequency item data and the wind direction frequency item data corresponding to different time scales are processed respectively by a sliding window rolling method to obtain multiple groups of wind speed training data and multiple groups of wind direction training data of different frequency items corresponding to different time scales, including:

[0026] Set the sliding window step size;

[0027] Slide the sliding window step on the wind speed frequency item data at different time scales until all the wind speed frequency item data at all time scales are traversed to obtain multiple groups of wind speed training data of different frequency items corresponding to different time scales;

[0028] The sliding window step is slid on the wind direction frequency item data at different time scales until the wind direction frequency item data at all time scales are traversed, so as to obtain multiple groups of wind direction training data of different frequency items corresponding to different time scales.

[0029] In some embodiments of the present invention, the convolutional neural network, the long short-term memory network and the self-attention mechanism layer are used to perform model training and construction on the multiple groups of wind speed training data and the multiple groups of wind direction training data of different frequency items corresponding to different time scales to obtain a wind speed and direction prediction model, including:

[0030] Normalizing the multiple groups of wind speed training data and the multiple groups of wind direction training data of different frequency items corresponding to different time scales, respectively, to obtain multiple groups of standard wind speed data and multiple groups of standard wind direction data of different frequency items corresponding to different time scales;

[0031] The convolutional neural network is used to extract and expand the features of the multiple groups of standard wind speed data and the multiple groups of standard wind direction data of different frequency items corresponding to different time scales, respectively, to obtain one-dimensional time series data of wind speed and one-dimensional time series data of wind direction of different frequency items corresponding to different time scales;

[0032] The one-dimensional time series data of wind speed and one-dimensional time series data of wind direction of different frequency items corresponding to different time scales are predicted respectively by the long short-term memory network to obtain first predicted data of wind speed and first predicted data of wind direction of different frequency items corresponding to different time scales;

[0033] The first wind speed prediction data and the first wind direction prediction data of different frequency items corresponding to different time scales are subjected to feature learning through the self-attention mechanism layer to obtain the second wind speed prediction data and the second wind direction prediction data of different frequency items corresponding to different time scales;

[0034] Denormalizing the second wind speed prediction data and the second wind direction prediction data of different frequency items corresponding to different time scales and then adding them together to obtain predicted average wind speed and predicted average wind direction of different time scales;

[0035] A wind speed and direction prediction model is constructed based on the predicted average wind speed and the predicted average wind direction at different time scales.

[0036] In some embodiments of the present invention, the step of substituting the current cabin rotation angular velocity and the current cabin heading angle into the MPC predictive control model for prediction and optimization based on a preset time step to obtain a target yaw angular velocity and a target yaw angle includes:

[0037] Based on a preset time step, the current cabin rotation angular velocity and the current cabin heading angle are substituted into the MPC predictive control model for prediction to obtain a predicted state sequence; wherein the predicted state sequence includes a plurality of predicted state data corresponding to the preset time step, each of the predicted state data includes a predicted value of the cabin rotation angular velocity and heading angle in the next period under an ideal state;

[0038] Performing evolutionary calculations according to the predicted state sequence and the MPC predictive control model to obtain a plurality of control input sequences corresponding to the predicted state sequence; wherein the control input sequence comprises a plurality of control input data corresponding to a preset time step, and each of the control input data comprises an angular velocity and a heading angle of rotation that the cabin needs to take in the next period under an actual state;

[0039] Performing optimization calculation according to the multiple control input sequences and a preset cost function, and determining the control input sequence that minimizes the cost function value as the target control input sequence;

[0040] The cabin rotation angular velocity corresponding to the first control input data in the target control input sequence is determined as the target yaw angular velocity, and the cabin heading angle corresponding to the first control input data in the target control input sequence is determined as the target yaw angle.

[0041] In a second aspect, an embodiment of the present invention provides an offshore wind turbine yaw control device based on MPC, comprising at least one control processor and a memory for communicating with the at least one control processor; the memory stores instructions executable by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute the offshore wind turbine yaw control method based on MPC as described in the first aspect above.

[0042] In a third aspect, an embodiment of the present invention provides an electronic device, comprising the MPC-based offshore wind turbine yaw control device as described in the second aspect above.

[0043] In a fourth aspect, an embodiment of the present invention provides a computer-readable storage medium storing computer-executable instructions, wherein the computer-executable instructions are used to execute the MPC-based offshore wind turbine yaw control method as described in the first aspect above.

[0044] The MPC-based offshore wind turbine yaw control method according to an embodiment of the present invention has at least the following beneficial effects:

[0045] Acquire multiple groups of historical wind speed data and multiple groups of historical wind direction data; perform multi-time scale average processing on the multiple groups of historical wind speed data and multiple groups of historical wind direction data, and then perform variational mode decomposition to obtain multiple groups of wind speed training data and multiple groups of wind direction training data with different frequency items corresponding to different time scales; Based on convolutional neural networks, long short-term memory networks and self-attention mechanism layers, perform model training and construction on multiple groups of wind speed training data and multiple groups of wind direction training data with different frequency items corresponding to different time scales to obtain a wind speed and wind direction prediction model; According to the current wind speed and wind speed and wind direction prediction model, predict the average wind speed of the next period; According to the current wind direction and A wind speed and direction prediction model is used to predict the average wind direction angle of the next period; a model is trained and constructed based on the average wind speed of the next period, the average wind direction angle of the next period, and the orientation angle of the offshore wind turbine cabin at the current moment to obtain an MPC prediction control model; based on a preset time step, the current cabin rotation angular velocity and the current cabin orientation angle are substituted into the MPC prediction control model for prediction and optimization to obtain a target yaw angular velocity and a target yaw angle; based on the target yaw angular velocity and the target yaw angle, the angular velocity of the offshore wind turbine is adjusted to the target yaw angular velocity, and the orientation angle of the offshore wind turbine is adjusted to the target yaw angle. According to the technical solution of an embodiment of the present invention, by constructing a VMD-SA-CNN-LSTM wind speed and direction prediction model and an MPC prediction control model, it is possible to accurately predict wind speed and wind direction, quickly respond to changes in wind direction, and improve the accuracy of yaw control. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] Figure 1 is a flow chart of an offshore wind turbine yaw control method based on MPC provided by an embodiment of the present invention;

[0047] Figure 2 yes Figure 1 Flow chart of step S12;

[0048] Figure 3 yes Figure 2 Flow chart of step S21;

[0049] Figure 4 yes Figure 2 Flow chart of step S23;

[0050] Figure 5 yes Figure 1 Flow chart of step S13;

[0051] Figure 6 This is the structure diagram of the VMD-SA-CNN-LSTM wind speed and direction prediction model;

[0052] Figure 7 yes Figure 1 Flow chart of step S16;

[0053] Figure 8 It is a structural diagram of an offshore wind turbine yaw control device based on MPC provided in another embodiment of the present invention. DETAILED DESCRIPTION

[0054] Embodiments of the present invention are described in detail below, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and cannot be understood as limiting the present invention.

[0055] In the description of the present invention, it should be understood that descriptions involving orientations, such as up, down, front, back, left, right, etc., and orientations or positional relationships indicated are based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as a limitation on the present invention.

[0056] In the description of the present invention, "several" means one or more, "more" means more than two, "greater than", "less than", "exceed" etc. are understood as not including the number itself, and "above", "below", "within" etc. are understood as including the number itself. If there is a description of "first" or "second", it is only used for the purpose of distinguishing the technical features, and cannot be understood as indicating or implying the relative importance or implicitly indicating the number of the indicated technical features or implicitly indicating the order of the indicated technical features.

[0057] In the description of the present invention, unless otherwise clearly defined, terms such as setting, installing, connecting, etc. should be understood in a broad sense, and technicians in the relevant technical field can reasonably determine the specific meanings of the above terms in the present invention in combination with the specific content of the technical solution.

[0058] The embodiment of the present invention provides an offshore wind turbine yaw control method based on MPC, an offshore wind turbine yaw control device based on MPC, an offshore wind turbine yaw control device based on MPC, and a computer-readable storage medium. The offshore wind turbine yaw control method based on MPC includes: performing multi-time scale average processing and variational mode decomposition on the acquired historical wind speed and wind direction data, and then training the wind speed and wind direction training data based on a convolutional neural network, a long short-term memory network, and a self-attention mechanism layer to construct a wind speed and wind direction prediction model; predicting the average wind speed and wind direction of the next period of time based on the current wind speed and wind direction and the wind speed and wind direction prediction model; training the predicted average wind speed and wind direction and the current cabin angle to construct an MPC prediction control model; substituting the current angular velocity and angle into the prediction control model for prediction and optimization to obtain the target yaw angular velocity and target yaw angle; adjusting the angular velocity to the target yaw angular velocity, and adjusting the heading angle to the target yaw angle; the method can accurately predict the wind speed and wind direction, quickly respond to wind direction changes, and improve the accuracy of yaw control.

[0059] The following further describes the MPC-based offshore wind turbine yaw control method according to an embodiment of the present invention based on the accompanying drawings.

[0060] Reference Figure 1 , Figure 1 A flowchart of a yaw control method for an offshore wind turbine generator system based on MPC is provided in an embodiment of the present invention. The yaw control method for an offshore wind turbine generator system based on MPC includes but is not limited to the following steps:

[0061] S11, obtaining multiple sets of historical wind speed data and multiple sets of historical wind direction data;

[0062] S12, for the multiple sets of historical wind speed data and the multiple sets of historical wind direction data, first perform multi-time scale average processing, and then perform variational mode decomposition to obtain multiple sets of wind speed training data and multiple sets of wind direction training data with different frequency items corresponding to different time scales;

[0063] S13, based on the convolutional neural network, the long short-term memory network and the self-attention mechanism layer, the model training and construction are performed on multiple groups of wind speed training data and multiple groups of wind direction training data with different frequency items corresponding to different time scales to obtain the wind speed and direction prediction model;

[0064] S14, predicting the average wind speed of the next period according to the current wind speed and wind speed and direction prediction model; predicting the average wind direction angle of the next period according to the current wind direction and wind speed and direction prediction model;

[0065] S15, training and constructing a model according to the average wind speed of the next period, the average wind direction angle of the next period, and the orientation angle of the nacelle of the offshore wind turbine at the current moment, to obtain an MPC predictive control model;

[0066] S16, based on a preset time step, substituting the current cabin rotation angular velocity and the current cabin heading angle into the MPC predictive control model for prediction and optimization to obtain a target yaw angular velocity and a target yaw angle;

[0067] S17, according to the target yaw angular velocity and the target yaw angle, adjusting the angular velocity of the offshore wind turbine to the target yaw angular velocity, and adjusting the heading angle of the offshore wind turbine to the target yaw angle.

[0068] In step S11, multiple sets of historical wind speed data and multiple sets of historical wind direction data are obtained, and the obtained data need to be preprocessed. The preprocessing steps include but are not limited to removing duplicate data, abnormal data or missing data, converting the data into a format suitable for subsequent analysis and processing, and standardizing the data.

[0069] In step S12, multi-time scale average processing refers to dividing the data according to different time scales, and then calculating the average value of all data on each divided time scale. Specifically, multiple groups of historical wind speed data and historical wind direction data are divided and averaged according to 10 seconds, 20 seconds, and 30 seconds time scales, respectively.

[0070] More specifically, the calculation formula for multi-time scale average processing of multiple sets of historical wind speed data and multiple sets of historical wind direction data is:

[0071]

[0072] in, It is expressed as the N-second average wind speed or average wind direction data at the i-th moment. Expressed as the average wind speed or average wind direction data j seconds before the i-th moment; i and j are positive integers; the average wind speed N is 10; the average wind direction N is 10, 20 or 30.

[0073] In step S12, variational modal decomposition (VMD) can decompose the multi-component signal into multiple single-component amplitude-frequency modulation signals at one time. In a specific embodiment, VMD is used to process multiple groups of average wind speed data and average wind direction data of different time scales, and decompose them into multiple component data, namely wind speed frequency item data and wind direction frequency item data.

[0074] Specifically, for each time scale (for example, 10 seconds), multiple groups of average wind speed data will be decomposed by VMD into three frequency items: wind speed high-frequency item data, wind speed medium-frequency item data, and wind speed low-frequency item data. Among them, the wind speed residual item data is the frequency data remaining after the high-frequency, medium-frequency and low-frequency components are extracted from the average wind speed data, that is, the result after the average wind speed data is subtracted from its wind speed high-frequency item data, wind speed medium-frequency item data and wind speed low-frequency item data. Similarly, multiple groups of average wind direction data will also be decomposed by VMD into three frequency items at the same time scale: wind direction high-frequency item data, wind direction medium-frequency item data, and wind direction low-frequency item data. The wind direction residual item data is the remaining part of the average wind direction data after the high-frequency, medium-frequency and low-frequency components are extracted, that is, the result after the average wind direction data is subtracted from its wind direction high-frequency item data, wind direction medium-frequency item data and wind direction low-frequency item data.

[0075] More specifically, in the variational mode decomposition process, multiple groups of average wind speed data and multiple groups of average wind direction data are decomposed into k intrinsic mode functions, respectively. The variational mode decomposition formula is:

[0076]

[0077] Wherein, the value of k is a positive integer, which represents the number of types of frequency item data. In one embodiment, the value of k is 3; k is the intrinsic mode function of each modal function, namely, the high-frequency item data of wind speed, the medium-frequency item data of wind speed, the low-frequency item data of wind speed, and the high-frequency item data of wind direction, the medium-frequency item data of wind direction, and the low-frequency item data of wind direction; ω k is the center frequency of each intrinsic mode function; δ(t) is the Dirac function; f is the average wind speed data or the average wind direction data; the wind speed residual term data is the residual frequency data after subtracting the extracted wind speed high-frequency term data, wind speed medium-frequency term data, and wind speed low-frequency term data from the average wind speed data; the wind direction residual term data is the residual frequency data after subtracting the extracted wind direction high-frequency term data, wind direction medium-frequency term data, and wind direction low-frequency term data from the average wind direction data.

[0078] In order to solve the variational mode decomposition formula, the augmented Lagrangian function is introduced to transform the constrained variational problem into an unconstrained variational problem. The calculation expression is:

[0079]

[0080] Among them, α is the penalty factor and λ is the Lagrange multiplier. According to the calculation expression, the Lagrange multiplier λ is iteratively updated to obtain the optimal solution u k 、w k , we can get the high-frequency item data, medium-frequency item data, low-frequency item data and residual item data after the average wind speed data and average wind direction data are decomposed respectively.

[0081] In step S13, the wind speed frequency item data and wind direction frequency item data are separated by variational mode decomposition, normalized first, and then the convolutional neural network (CNN) is used to extract features from each frequency item data. The time series is predicted by the long short-term memory network (LSTM), and the self-attention mechanism layer (SA) is combined to perform dynamic feature learning. Finally, the predicted frequency items and residual items are output in turn, and the predicted average wind speed and predicted average wind direction corresponding to each frequency item are obtained by adding them after denormalization.

[0082] It should be noted that each set of wind speed frequency item data and each set of wind direction frequency item data should be based on the convolutional neural network, the long short-term memory network, and the self-attention mechanism layer for data training, output the prediction results of the corresponding frequency items, and construct a wind speed and direction prediction model; that is, the wind speed high-frequency item data, wind speed medium-frequency item data, wind speed low-frequency item data, wind speed residual item data and the wind direction high-frequency item data, wind direction medium-frequency item data, wind direction low-frequency item data, and wind direction residual item data should all be input into the convolutional neural network, the long short-term memory network, and the self-attention mechanism layer in sequence for data training to obtain the predicted average wind speed and predicted average wind direction.

[0083] It should be noted that the wind speed and direction prediction model is a prediction model based on variational mode decomposition (VMD), convolutional neural network (CNN), long short-term memory network (LSTM), and self-attention mechanism layer (SA), that is, the VMD-SA-CNN-LSTM wind speed and direction prediction model.

[0084] In step S14, the current wind speed and current wind direction are generally measured by a physical sensor for measuring wind speed and wind direction, such as a wind vane. The measured current wind speed is substituted into the wind speed and wind direction prediction model to predict the average wind speed for the next period; the measured current wind direction is substituted into the wind speed and wind direction prediction model to predict the average wind direction angle for the next period.

[0085] In step S15 to step S16, the average wind speed and the average wind direction of the next period are predicted by the VMD-SA-CNN-LSTM wind speed and direction prediction model; the orientation angle of the offshore wind turbine cabin at the current moment is given by the cabin sensor; the yaw error refers to the angle between the current orientation of the cabin and the ideal orientation. The yaw speed refers to the angular velocity of the cabin; the yaw direction refers to the orientation angle of the cabin.

[0086] The predicted state sequence refers to the prediction of the yaw state of the wind turbine in the future, including multiple predicted state data corresponding to the preset time step. Each predicted state data includes the predicted values ​​of the rotation angular velocity and heading angle of the nacelle in the next period of time under ideal conditions.

[0087] The control input sequence refers to a series of instructions or signals planned to adjust the yaw angle based on the current nacelle heading angle and the predicted yaw state of the wind turbine in the future. It contains multiple control input data corresponding to the preset time step, and each control input data includes the rotation angular velocity and heading angle that the nacelle needs to take in the next period of time under the actual state.

[0088] The preset time step refers to the length of time predicted by the MPC (Model Predictive Control) predictive control model. In one embodiment, the MPC predictive control model predicts the next 10 time steps, each time step is 1 second, that is, it predicts the target yaw velocity and target yaw angle within the next 10 seconds. That is to say, in the predicted state sequence, there are 10 predicted state data for the next 10 seconds, and each state data contains the predicted values ​​of the rotation angular velocity and heading angle of the cabin within 1 second under ideal conditions; in the control input sequence, there are 10 control input data for the next 10 seconds, and each control input data contains the rotation angular velocity and heading angle that the cabin needs to take within 1 second under actual conditions. It should be noted that the time step can be set according to actual needs.

[0089] In the MPC predictive control model, the average wind speed and the average wind direction angle of the next period can be predicted by the VMD-SA-CNN-LSTM wind speed and wind direction prediction model, as well as the orientation angle of the offshore wind turbine cabin at the current moment, to obtain the predicted state sequence, that is, the rotation angular velocity and orientation angle of the cabin in the next period under ideal conditions, and then calculate the yaw error. Based on the yaw error, the MPC predictive control model considers multiple influencing factors such as the square of the yaw error, the average wind speed, the yaw action distance, the yaw action time, the number of yaw actions, and the derivative of the yaw speed for optimization, and evolves according to multiple possible influencing factors to obtain multiple control input sequences, and then finds the control input sequence that meets the constraints from the multiple control input sequences as the target control sequence.

[0090] In the MPC predictive control model, the orientation angle of the cabin predicted at the next moment under ideal conditions is expressed as:

[0091]

[0092] Among them, θ np(k+1) is the predicted orientation angle of the cabin at the next moment; θ np (k) is the current cabin heading angle; To predict the angular velocity of the cabin in the next period; T c is time; when T c When it is 1 second, the predicted direction angle of the cabin in the next second.

[0093] When T c When is 1 second, the yaw error of the next second is expressed as:

[0094]

[0095] Among them, θ ye (k+1|k) is the yaw error of the next second; θ wd (k+1|k) is the predicted average wind direction for the next second; θ np (k) is the current cabin heading angle; To predict the angular velocity of the nacelle in the next period of time.

[0096] Specifically, when the preset time step is 10 and each time step is 1 second, the MPC predictive control model is expressed as:

[0097]

[0098]

[0099] Among them, the time step of the MPC predictive control model is 10.

[0100] (k+1|k) represents the cabin state at the k+1th moment predicted at the kth moment;

[0101] (k+2|k) represents the cabin state at the k+2th moment predicted by the kth moment;

[0102] (k+3|k) represents the cabin state at the k+3th moment predicted by the kth moment;

[0103] (k+4|k) represents the cabin state at the k+4th moment predicted by the kth moment;

[0104] (k+5|k) represents the cabin state at the k+5th moment predicted by the kth moment;

[0105] (k+6|k) represents the cabin state at the k+6th moment predicted by the kth moment;

[0106] (k+7|k) represents the cabin state at the k+7th moment predicted by the kth moment;

[0107] (k+8|k) represents the cabin state at the k+8th moment predicted by the kth moment;

[0108] (k+9|k) represents the cabin state at the k+9th moment predicted by the kth moment;

[0109] (k+10|k) represents the cabin state at the k+10th moment predicted at the kth moment.

[0110] It should be noted that the MPC predictive control model optimization process is to find the control input sequence corresponding to the minimum cost function value from multiple control input sequences as the target control sequence, and then apply the target yaw velocity and target yaw angle represented by the target input sequence to the next prediction process.

[0111] The constraints of the MPC predictive control model, that is, the cost function, are expressed as:

[0112]

[0113]

[0114] It should be noted that QF(1) takes into account the influence of the yaw error on the output power of the wind turbine. Since the yaw error directly affects the output power in the form of the square of the cosine, QF(1) is expressed in the form of the square of the yaw error.

[0115] It should be noted that when considering the influence of wind speed on the yaw error of wind turbines, QF(2) divides the operating state of the wind turbine into three different intervals: interval 1 covers the wind speed range of [3,5] or [12,24], interval 2 covers the wind speed range of [5,9], and interval 3 covers the wind speed range of [9,12]. Based on these wind speed intervals, the parameter ω2 in QF(2) will be dynamically adjusted to ω 21 ,ω 22 or 23 , to adapt to different wind speed conditions; in addition, each wind speed range corresponds to a different yaw control target: when the wind speed exceeds the rated wind speed of 12m / s, reduce the yaw action to limit power; when the wind speed is 9m / s to 12m / s, reduce the yaw error to optimize power output; when the wind speed is 5m / s to 9m / s, balance the yaw accuracy and number; when the wind speed is 3m / s to 5m / s, pay attention to the use of the yaw actuator; when the wind speed is lower than the cut-in wind speed of 3m / s or higher than the cut-out wind speed of 24m / s, the wind turbine is shut down.

[0116] It should be noted that QF(3) takes into account the yaw distance of the wind turbine and the sampling time T c By minimizing the yaw distance, the wind turbine is always kept in the optimal direction, improving energy efficiency and performance.

[0117] It should be noted that QF (4) takes into account the yaw time of the wind turbine. When the yaw speed of the wind turbine is not zero, the yaw time is accumulated by one second each time. Increasing the yaw time will lead to increased energy consumption and wear of the nacelle. Minimizing the yaw time helps to improve the long-term stability of the wind turbine.

[0118] It should be noted that QF (5) takes into account the number of wind turbine starts and stops. When the yaw speed changes from zero to non-zero or from non-zero to zero, the number of wind turbine starts and stops increases once. Frequent starts and stops will cause premature wear of the wind turbine and increase maintenance costs. By minimizing the number of starts and stops, the service life of the wind turbine can be extended and the economic benefits can be improved.

[0119] It should be noted that QF (6) takes into account the derivative of the yaw speed of the wind turbine to avoid excessive yaw speed of the wind turbine; by controlling the rate of change of the yaw speed, violent movement of the wind turbine can be avoided, thereby reducing the load and energy consumption of the equipment.

[0120] In step S17, a yaw action is performed according to the target yaw angular velocity and the target yaw angle, that is, the angular velocity of the offshore wind turbine is adjusted to the target yaw angular velocity, and the heading angle of the offshore wind turbine is adjusted to the target yaw angle.

[0121] Through step S11 to step S17, by constructing a VMD-SA-CNN-LSTM wind speed and wind direction prediction model, the average wind speed and the average wind direction angle of the next period are predicted, and the model training and construction are carried out according to the predicted average wind speed and the average wind direction angle of the next period and the orientation angle of the offshore wind turbine cabin at the current moment, and an MPC prediction control model is constructed; the rotation angular velocity of the cabin in the next period is predicted according to the MPC prediction control model, and the yaw error is calculated; the control input sequence is optimized by considering various influencing factors such as the yaw error, and the target yaw angular velocity and the target yaw angle in the target control input sequence are obtained; according to the target yaw angular velocity and the target yaw angle, the angular velocity of the offshore wind turbine is adjusted to the target yaw angular velocity, and the orientation angle of the offshore wind turbine is adjusted to the target yaw angle. This method can accurately predict wind speed and wind direction, quickly respond to wind direction changes, and improve the accuracy of yaw control.

[0122] According to some embodiments of the present invention, referring to Figure 2 ,exist Figure 1 Step S12 of the illustrated embodiment also includes but is not limited to the following steps:

[0123] S21, performing multi-time scale average processing on the multiple sets of historical wind speed data and the multiple sets of historical wind direction data, respectively, to obtain multiple sets of average wind speed data and multiple sets of average wind direction data corresponding to different time scales;

[0124] S22, calculating multiple groups of average wind speed data and multiple groups of average wind direction data corresponding to different time scales based on a variational mode decomposition formula to obtain wind speed frequency item data and wind direction frequency item data corresponding to different time scales;

[0125] S23, the wind speed frequency item data and the wind direction frequency item data corresponding to different time scales are processed by a sliding window rolling method to obtain multiple groups of wind speed training data and multiple groups of wind direction training data of different frequency items corresponding to different time scales.

[0126] Through steps S21 to S23, multi-time scale average processing can remove some high-frequency noise or fluctuations in the data, improve the stability of the wind speed and direction prediction model, and improve the prediction accuracy. The variational mode decomposition method can capture multiple frequency components in multiple groups of average wind speed data and multiple groups of average wind direction data, obtain complex features in wind speed and direction, and improve the prediction accuracy of the wind speed and direction prediction model.

[0127] According to some embodiments of the present invention, referring to Figure 3 ,exist Figure 2 In step S21 of the illustrated embodiment, the following steps are also included but not limited to:

[0128] S31, dividing the multiple groups of historical wind speed data and the multiple groups of historical wind direction data according to different time scales, respectively, to obtain multiple groups of first wind speed data and multiple groups of first wind direction data corresponding to different time scales;

[0129] S32, performing averaging processing on multiple groups of first wind speed data corresponding to different time scales to obtain multiple groups of average wind speed data corresponding to different time scales;

[0130] S33, performing averaging processing on multiple groups of first wind direction data corresponding to different time scales to obtain multiple groups of average wind direction data corresponding to different time scales.

[0131] By dividing the time scales into different ones through steps S31 to S33, we can better understand the changing trends and periodicity of wind speed and wind direction in different time ranges. Averaging can remove some high-frequency noise or fluctuations in the data, thereby improving the data quality.

[0132] According to some embodiments of the present invention, referring to Figure 4 ,exist Figure 2 Step S23 of the illustrated embodiment also includes but is not limited to the following steps:

[0133] S41, setting the sliding window step size;

[0134] S42, sliding the sliding window step size on the wind speed frequency item data at different time scales until all wind speed frequency item data at all time scales are traversed to obtain multiple groups of wind speed training data of different frequency items corresponding to different time scales;

[0135] S43, sliding the sliding window step on the wind direction frequency item data at different time scales until the wind direction frequency item data at all time scales are traversed, and obtaining multiple groups of wind direction training data of different frequency items corresponding to different time scales.

[0136] It should be noted that, in one embodiment, the step size of the sliding window is set to 1. This sliding window will move on the wind speed frequency data of various time scales in turn until the wind speed frequency data at all time scales are covered and processed. In this process, multiple groups of wind speed training data can be obtained, which correspond to different time scales and frequency items. Similarly, the sliding window will also move on the wind direction frequency data with the same step size until the wind direction frequency data at all time scales are traversed, and multiple groups of wind direction training data corresponding to different time scales and frequency items are obtained. Each group of wind speed training data and each group of wind direction training data are respectively organized in the form of a one-dimensional matrix, and the matrix contains the input data of 40 sampling points. For wind speed training data, each one-dimensional array contains wind speed measurement values ​​at 40 consecutive time points; similarly, for wind direction training data, each one-dimensional array contains wind direction measurement values ​​at 40 consecutive time points.

[0137] Through steps S41 to S43, the sliding window technology is used to fully utilize the information in the wind speed frequency item data and the wind direction frequency item data, extract the wind speed and wind direction characteristics at different time scales, and at the same time consider the data characteristics of different time scales and frequency items, so as to more accurately capture the complexity of wind speed and wind direction.

[0138] According to some embodiments of the present invention, referring to Figures 5 and 6 ,exist Figure 1 Step S13 of the illustrated embodiment also includes but is not limited to the following steps:

[0139] S51, respectively normalizing the multiple groups of wind speed training data and the multiple groups of wind direction training data of different frequency items corresponding to different time scales to obtain multiple groups of standard wind speed data and multiple groups of standard wind direction data of different frequency items corresponding to different time scales;

[0140] S52, extracting and expanding features of multiple groups of standard wind speed data and multiple groups of standard wind direction data of different frequency items corresponding to different time scales by using a convolutional neural network, to obtain one-dimensional time series data of wind speed and one-dimensional time series data of wind direction of different frequency items corresponding to different time scales;

[0141] S53, predicting the one-dimensional time series data of wind speed and the one-dimensional time series data of wind direction of different frequency items corresponding to different time scales respectively through a long short-term memory network to obtain first predicted wind speed data and first predicted wind direction data of different frequency items corresponding to different time scales;

[0142] S54, performing feature learning on the first wind speed prediction data and the first wind direction prediction data of different frequency items corresponding to different time scales through a self-attention mechanism layer to obtain the second wind speed prediction data and the second wind direction prediction data of different frequency items corresponding to different time scales;

[0143] S55, performing denormalization on the second wind speed prediction data and the second wind direction prediction data of different frequency items corresponding to different time scales and then adding them together to obtain the predicted average wind speed and the predicted average wind direction of different time scales;

[0144] S56: construct a wind speed and direction prediction model according to the predicted average wind speed and predicted average wind direction at different time scales.

[0145] In step S51, the calculation formula for normalization processing is:

[0146]

[0147] train norm =(train-train mean ) / train std ;

[0148] Among them, A pq is the data of the pth row and qth column of the training data matrix, R is the total number of rows in the training data matrix, train q-mean is the average value of the average wind direction data in the qth column, train q-std is the standard value of the average wind direction data in the qth column; train norm is the normalized standard wind speed data.

[0149] In step S52, the convolution layer includes convolution and pooling operations. In the convolution operation, each convolution kernel slides on the input data (such as standard wind speed data or standard wind direction data), covering a small part of the data (local area) each time, and calculating the weighted sum of the area and the convolution kernel, thereby extracting the features of the area. The extracted features are expressed as:

[0150]

[0151] Among them, Y(i,j) is the feature extracted after convolution; X is the input time series; W is the convolution kernel weight matrix; k is the convolution kernel size; m and n are the weight positions inside the convolution kernel; b is the bias term.

[0152] The pooling operation reduces the size of the convolutional feature map by selecting the maximum value within the pooling window while retaining the salient features of the data, expressed as:

[0153]

[0154] Among them, P(i,j) is the output feature after pooling; A is the feature map after the convolution activation function; s is the stride, which indicates the step size of each sliding of the pooling window; m and n are the sizes of the pooling window.

[0155] In one embodiment, the features of the input standard wind speed data and standard wind direction data are first extracted through the convolution layer using 10 convolution kernels of size 2×1. This process produces 10 feature maps, each of which has a size of 39x1, which effectively capture the local features of the data. Subsequently, the pooling layer adopts the maximum pooling strategy and samples these feature maps using a pooling window of size 2×1 (with a step size of 1); after processing by the pooling layer, the feature data will be further passed to the flattening layer and converted into a 390×1 one-dimensional time series data, which is then input into the long short-term memory network.

[0156] In step S53, the long short-term memory network receives the one-dimensional time series data of the flattened layer, uses the gate mechanism and the cell state update formula to learn the long-term dependencies in the data, uses its time series modeling ability to predict the input time series, and outputs the prediction results as input to the self-attention mechanism layer.

[0157] The LSTM network includes three gates (input gate, forget gate, output gate) and a cell state, which enables the LSTM unit to store useful information for a long time and capture long-term dependencies. The forget gate determines the information deleted from the cell state in the previous time step, the input gate controls the information received by the cell state, and the output gate controls the output information of the cell state; the information formula output by each gating mechanism is as follows:

[0158] f t =σ(W f [h t-1 ,x t ]+b f );

[0159] i t =σ(W i [h t-1 ,x t ]+b i );

[0160] o t =σ(W o [h t-1 ,xt ]+b o );

[0161] Among them, W f is the forget gate and the input x t , the hidden state h of the previous time step t-1 The weight matrix between i is the input gate and input x t , the hidden state h of the previous time step t-1 The weight matrix between o is the output gate and input x t , the hidden state h of the previous time step t-1 The weight matrix between f , b i , b o are the corresponding biases respectively; f t is the output of the forget gate; i t is the input gate output; o t is the output of the output gate; σ is the sigmoid activation function, which converts the output to the interval [0,1].

[0162] In one embodiment, the LSTM network layer is composed of 32 LSTM units. For each input time step, the layer generates a 32-dimensional output vector. The LSTM network performs deep processing on the input one-dimensional time series data through its unique gating mechanism (including input gate, forget gate and output gate) and cell state to make predictions. The prediction results are expressed as multiple 32×1 matrices X, which contain the LSTM network's understanding and prediction information of the time series data; subsequently, these prediction results are sent to the self-attention mechanism layer for further analysis and processing.

[0163] In step S54, the self-attention mechanism layer outputs predicted data based on the long short-term memory network, utilizes the correlation between keys, queries, and values, calculates the importance of each part through the weighted distribution principle, and dynamically adjusts the weights to achieve accurate attention and efficient processing of input information.

[0164] In one embodiment, the self-attention layer includes a single-head self-attention mechanism and a fully connected layer, that is, the value of the layer head of the self-attention mechanism layer is set to 1, and the value of the key is set to 2; the self-attention mechanism uses the correlation between the key, query and value to calculate the importance of each part through the weighted distribution principle, and the attention weight matrix assigns weights to each element of the value matrix. The higher the weight, the more important the element. The size of the fully connected layer depends on the time scale of its averaging. When the time scale is 10 seconds of average wind direction, the fully connected layer has 10 neurons; when the time scale is 20 seconds of average wind direction, the fully connected layer has 20 neurons; when the time scale is 30 seconds of average wind direction, the fully connected layer has 30 neurons. The fully connected layer integrates all learned features and reshapes the shape of the output matrix, and finally connects the regression layer to return the predicted output data, the size of which depends on the time scale of its averaging.

[0165] Q=XW Q +b Q ;

[0166] K=XW K +b K ;

[0167] V=XW V +b V ;

[0168] Among them, Q, K, V are query matrix, key matrix and value matrix respectively; W Q , W K , W V is the corresponding weight matrix, the size is 2×32; X is the input time series, which is a 32x1 input matrix. Q 、b K 、b V is the corresponding bias matrix, and its size is 2×1.

[0169] Through dot product and scaling operations, we get the attention score matrix a n , use the Softmax function to calculate the above attention score a n Normalize and get the weight coefficient Finally, the weight coefficient The weighted sum is taken with the value vector V to obtain the final output I;

[0170]

[0171] Among them, n and N represent the attention score matrix The dimension of I is 32×1 output matrix; K T is the transposed matrix of K, d k is the scaling factor.

[0172] It should be noted that by introducing R 2 The coefficient, MAE and RMSE evaluation indicators quantitatively measure the prediction accuracy of the VMD-SA-CNN-LSTM wind speed and direction prediction model.

[0173] R 2 The coefficient is used to measure the fit of the model, R 2 Close to 1 means the model fits well and the predicted data can fit the real data well. 2 Close to 0 means the model has weak predictive ability, and the calculation expression is:

[0174]

[0175] Among them, y i is the true value; is the predicted value; is the average of the true values; n is the total number of data points.

[0176] MAE is the average of the absolute difference between the true value and the predicted value, which is used to measure the size of the model prediction error. The smaller the MAE, the higher the prediction accuracy of the model. The calculation expression is:

[0177]

[0178] Among them, y i is the true value; is the predicted value; n is the total number of data points.

[0179] RMSE is the square root of the average of the squared errors between the predicted value and the true value. It is used to measure the prediction error of the model and is particularly sensitive to larger errors. The calculation expression is:

[0180]

[0181] In one embodiment, to further evaluate the performance of the yaw control method, the accuracy of wind direction tracking, the use of yaw actuators, and power generation are evaluated. The accuracy of wind direction tracking is quantitatively analyzed by the yaw error accumulation, average yaw error, and root mean square yaw error indicators, and the calculation expression is:

[0182] θ ye =θ wd -θ np ;

[0183]

[0184]

[0185] Among them, θ ye is the yaw error, which is determined by the wind direction angle θwd Subtract the cabin heading angle θ np It is concluded that TE(θ ye ) is the total yaw error, which measures the cumulative size of the yaw error; MAE(θ ye ) is the mean absolute error, reflecting the average level of yaw error; RMSE(θ ye ) is the root mean square error. Pay attention to the fluctuation of the yaw error, which is especially obvious when the yaw error is large.

[0186] Through steps S51 to S56, the VMD-SA-CNN-LSTM wind speed and direction prediction model achieves high-precision ultra-short-term prediction of wind speed and direction by giving full play to the advantages of CNN in local feature extraction, LSTM in time-dependent modeling, and SA in dynamic selection, thereby providing accurate wind speed and direction prediction data for yaw control.

[0187] According to some embodiments of the present invention, referring to Figure 6 ,exist Figure 1 Step S16 of the illustrated embodiment also includes but is not limited to the following steps:

[0188] S61, based on a preset time step, substitute the current cabin rotation angular velocity and the current cabin heading angle into the MPC predictive control model for prediction to obtain a predicted state sequence; wherein the predicted state sequence includes a plurality of predicted state data corresponding to the preset time step, each predicted state data including predicted values ​​of the cabin rotation angular velocity and heading angle in the next period under an ideal state;

[0189] S62, performing evolutionary calculations according to the predicted state sequence and the MPC predictive control model to obtain a plurality of control input sequences corresponding to the predicted state sequence; wherein the control input sequence includes a plurality of control input data corresponding to a preset time step, and each control input data includes a rotation angular velocity and a heading angle that the cabin needs to take in the next period under the actual state;

[0190] S63, performing optimization calculation according to the multiple control input sequences and the preset cost function, and determining the control input sequence that minimizes the cost function value as the target control input sequence;

[0191] S64, determining the cabin rotation angular velocity corresponding to the first control input data in the target control input sequence as the target yaw angular velocity, and determining the cabin heading angle corresponding to the first control input data in the target control input sequence as the target yaw angle.

[0192] It should be noted that, in the target control sequence corresponding to the minimum cost function value, only the first control input data will be applied, and the cabin rotation angular velocity corresponding to the first control input data in the target control input sequence is determined as the target yaw angular velocity, and the cabin heading angle corresponding to the first control input data in the target control input sequence is determined as the target yaw angle; then, based on the target yaw angular velocity, target yaw angle and MPC predictive control model, the target control sequence for the next time period is obtained for iterative optimization.

[0193] In step S61 to step S64, the MPC predictive control model can predict the state of the cabin in the future based on the current rotational angular velocity and heading angle of the cabin, and can understand the dynamic changes of the cabin in advance, so as to make more accurate control decisions; according to the predicted state sequence, multiple possible control input sequences are obtained through evolutionary calculation. These control input sequences represent the possible future states of the cabin under different control strategies. By selecting the control input sequence that minimizes the cost function as the target control input sequence, this process realizes the optimization of the control strategy and ensures that the cabin can operate smoothly and efficiently under actual conditions.

[0194] like Figure 8 As shown, Figure 8 : is a structural diagram of an offshore wind turbine yaw control device based on MPC provided by an embodiment of the present invention. The present invention also provides an offshore wind turbine yaw control device based on MPC, comprising:

[0195] The processor 801 may be implemented by a general-purpose central processing unit (CPU), a microprocessor, an application-specific integrated circuit (ASIC), or one or more integrated circuits, and is used to execute relevant programs to implement the technical solutions provided in the embodiments of the present application;

[0196] The memory 802 can be implemented in the form of a read-only memory (ROM), a static storage device, a dynamic storage device, or a random access memory (RAM). The memory 802 can store an operating system and other application programs. When the technical solution provided in the embodiment of this specification is implemented by software or firmware, the relevant program code is stored in the memory 802, and the processor 801 calls and executes the MPC-based offshore wind turbine yaw control method of the embodiment of this application;

[0197] Input / output interface 803, used to implement information input and output;

[0198] The communication interface 804 is used to realize the communication interaction between the present apparatus and other devices. The communication can be realized through a wired manner (such as USB, network cable, etc.) or a wireless manner (such as mobile network, WIFI, Bluetooth, etc.);

[0199] A bus 805 that transmits information between the various components of the device (e.g., the processor 801, the memory 802, the input / output interface 803, and the communication interface 804);

[0200] The processor 801 , the memory 802 , the input / output interface 803 and the communication interface 804 are connected to each other in communication within the device via a bus 805 .

[0201] An embodiment of the present application also provides an electronic device, including the above-mentioned MPC-based offshore wind turbine yaw control device.

[0202] An embodiment of the present application further provides a storage medium, which is a computer-readable storage medium and stores a computer program. When the computer program is executed by a processor, the above-mentioned MPC-based offshore wind turbine yaw control method is implemented.

[0203] As a non-transient computer-readable storage medium, the memory can be used to store non-transient software programs and non-transient computer executable programs. In addition, the memory may include a high-speed random access memory, and may also include a non-transient memory, such as at least one disk storage device, a flash memory device, or other non-transient solid-state storage devices. In some embodiments, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the processor via a network. Examples of the above-mentioned networks include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof. The device embodiments described above are merely schematic, wherein the units described as separate components may or may not be physically separated, and are implemented to be located in one place, or may also be distributed to multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the present embodiment.

[0204] It will be appreciated by those skilled in the art that all or some of the steps and systems in the disclosed method above may be implemented as software, firmware, hardware and appropriate combinations thereof. Some physical components or all physical components may be implemented as software executed by a processor, such as a central processing unit, a digital signal processor or a microprocessor, or may be implemented as hardware, or may be implemented as an integrated circuit, such as an application specific integrated circuit. Such software may be distributed on a computer-readable medium, which may include a computer storage medium (or a non-transitory medium) and a communication medium (or a temporary medium). As known to those skilled in the art, the term computer storage medium includes volatile and non-volatile, removable and non-removable media implemented in any method or technology for storing information (such as computer-readable instructions, data structures, program modules or other data). Computer storage media include, but are not limited to, RAM, ROM, EEPROM, flash memory or other memory technologies, CD-ROM, digital versatile disks (DVD) or other optical disk storage, magnetic cassettes, magnetic tapes, disk storage or other magnetic storage devices, or any other medium that may be used to store desired information and may be accessed by a computer. Furthermore, it is well known to those skilled in the art that communication media typically include computer readable instructions, data structures, program modules, or other data in a modulated data signal such as a carrier wave or other transport mechanism, and may include any information delivery media.

[0205] The above is a specific description of the preferred implementation of the present invention, but the present invention is not limited to the above-mentioned implementation mode. Technical personnel familiar with the field can also make various equivalent deformations or substitutions under the shared conditions without violating the spirit of the present invention. These equivalent deformations or substitutions are all included in the scope defined by the claims of the present invention.

Claims

1. A yaw control method for an offshore wind turbine based on MPC, characterized in that: include: Obtain multiple sets of historical wind speed data and multiple sets of historical wind direction data; The multiple sets of historical wind speed data and the multiple sets of historical wind direction data are respectively subjected to multi-time scale average processing and then subjected to variational mode decomposition to obtain multiple sets of wind speed training data and multiple sets of wind direction training data of different frequency items corresponding to different time scales; Based on a convolutional neural network, a long short-term memory network and a self-attention mechanism layer, model training and construction are performed on the multiple groups of wind speed training data and the multiple groups of wind direction training data of different frequency items corresponding to different time scales to obtain a wind speed and direction prediction model; According to the current wind speed and the wind speed and wind direction prediction model, the average wind speed of the next period is predicted; according to the current wind direction and the wind speed and wind direction prediction model, the average wind direction angle of the next period is predicted; Model training and construction are performed according to the average wind speed of the next period, the average wind direction angle of the next period, and the orientation angle of the offshore wind turbine cabin at the current moment to obtain an MPC predictive control model; Based on a preset time step, the current cabin rotation angular velocity and the current cabin heading angle are substituted into the MPC predictive control model for prediction and optimization to obtain a target yaw angular velocity and a target yaw angle; According to the target yaw angular velocity and the target yaw angle, the angular velocity of the offshore wind turbine generator set is adjusted to the target yaw angular velocity, and the orientation angle of the offshore wind turbine generator set is adjusted to the target yaw angle.

2. The MPC-based offshore wind turbine yaw control method according to claim 1, characterized in that: The multiple sets of historical wind speed data and the multiple sets of historical wind direction data are respectively subjected to multi-time scale average processing and then subjected to variational mode decomposition to obtain multiple sets of wind speed training data and multiple sets of wind direction training data with different frequency terms corresponding to different time scales, including: Performing multi-time scale average processing on the multiple sets of historical wind speed data and the multiple sets of historical wind direction data, respectively, to obtain multiple sets of average wind speed data and multiple sets of average wind direction data corresponding to different time scales; The multiple groups of average wind speed data and the multiple groups of average wind direction data corresponding to different time scales are calculated based on the variational mode decomposition formula to obtain wind speed frequency item data and wind direction frequency item data corresponding to different time scales; The wind speed frequency item data and the wind direction frequency item data corresponding to different time scales are processed respectively by a sliding window rolling method to obtain multiple groups of wind speed training data and multiple groups of wind direction training data of different frequency items corresponding to different time scales.

3. The MPC-based offshore wind turbine yaw control method according to claim 2, characterized in that: The multiple sets of historical wind speed data and the multiple sets of historical wind direction data are respectively subjected to multi-time scale average processing to obtain multiple sets of average wind speed data and multiple sets of average wind direction data corresponding to different time scales, including: Dividing the multiple groups of historical wind speed data and the multiple groups of historical wind direction data according to different time scales, respectively, to obtain multiple groups of first wind speed data and multiple groups of first wind direction data corresponding to different time scales; Averaging the multiple groups of first wind speed data corresponding to different time scales to obtain multiple groups of average wind speed data corresponding to different time scales; The multiple groups of first wind direction data corresponding to different time scales are averaged to obtain multiple groups of average wind direction data corresponding to different time scales.

4. The MPC-based offshore wind turbine yaw control method according to claim 2, characterized in that: The wind speed frequency item data includes wind speed high frequency item data, wind speed medium frequency item data, wind speed low frequency item data and wind speed residual item data; the wind direction frequency item data includes wind direction high frequency item data, wind direction medium frequency item data, wind direction low frequency item data and wind direction residual item data; The variational mode decomposition formula is: Among them, the value of k is a positive integer, which represents the number of types of frequency data; u k is the intrinsic mode function of each modal function, namely, the high-frequency item data of wind speed, the medium-frequency item data of wind speed, the low-frequency item data of wind speed, and the high-frequency item data of wind direction, the medium-frequency item data of wind direction, and the low-frequency item data of wind direction; ω k is the center frequency of each intrinsic mode function; δ(t) is the Dirac function; f is the average wind speed data or the average wind direction data; the wind speed residual term data is the residual frequency data after subtracting the extracted wind speed high-frequency term data, wind speed medium-frequency term data, and wind speed low-frequency term data from the average wind speed data; the wind direction residual term data is the residual frequency data after subtracting the extracted wind direction high-frequency term data, wind direction medium-frequency term data, and wind direction low-frequency term data from the average wind direction data.

5. The MPC-based offshore wind turbine yaw control method according to claim 2, characterized in that: The wind speed frequency item data and the wind direction frequency item data corresponding to different time scales are processed respectively by a sliding window rolling method to obtain multiple groups of wind speed training data and multiple groups of wind direction training data of different frequency items corresponding to different time scales, including: Set the sliding window step size; Slide the sliding window step on the wind speed frequency item data at different time scales until all the wind speed frequency item data at all time scales are traversed to obtain multiple groups of wind speed training data of different frequency items corresponding to different time scales; The sliding window step is slid on the wind direction frequency item data at different time scales until the wind direction frequency item data at all time scales are traversed, so as to obtain multiple groups of wind direction training data of different frequency items corresponding to different time scales.

6. The MPC-based offshore wind turbine yaw control method according to claim 1, characterized in that: The method is based on the convolutional neural network, the long short-term memory network and the self-attention mechanism layer, and performs model training and construction on the multiple groups of wind speed training data and the multiple groups of wind direction training data of different frequency items corresponding to different time scales to obtain a wind speed and direction prediction model, including: Normalizing the multiple groups of wind speed training data and the multiple groups of wind direction training data of different frequency items corresponding to different time scales, respectively, to obtain multiple groups of standard wind speed data and multiple groups of standard wind direction data of different frequency items corresponding to different time scales; The convolutional neural network is used to extract and expand the features of the multiple groups of standard wind speed data and the multiple groups of standard wind direction data of different frequency items corresponding to different time scales, respectively, to obtain one-dimensional time series data of wind speed and one-dimensional time series data of wind direction of different frequency items corresponding to different time scales; The one-dimensional time series data of wind speed and one-dimensional time series data of wind direction of different frequency items corresponding to different time scales are predicted respectively by the long short-term memory network to obtain first predicted data of wind speed and first predicted data of wind direction of different frequency items corresponding to different time scales; The first wind speed prediction data and the first wind direction prediction data of different frequency items corresponding to different time scales are subjected to feature learning through the self-attention mechanism layer to obtain the second wind speed prediction data and the second wind direction prediction data of different frequency items corresponding to different time scales; Denormalizing the second wind speed prediction data and the second wind direction prediction data of different frequency items corresponding to different time scales and then adding them together to obtain predicted average wind speed and predicted average wind direction of different time scales; A wind speed and direction prediction model is constructed based on the predicted average wind speed and the predicted average wind direction at different time scales.

7. The MPC-based offshore wind turbine yaw control method according to claim 1, characterized in that: Substituting the current cabin rotation angular velocity and the current cabin heading angle into the MPC predictive control model for prediction and optimization based on the preset time step to obtain the target yaw angular velocity and the target yaw angle includes: Based on a preset time step, the current cabin rotation angular velocity and the current cabin heading angle are substituted into the MPC predictive control model for prediction to obtain a predicted state sequence; wherein the predicted state sequence includes a plurality of predicted state data corresponding to the preset time step, each of the predicted state data includes a predicted value of the cabin rotation angular velocity and heading angle in the next period under an ideal state; Performing evolutionary calculations according to the predicted state sequence and the MPC predictive control model to obtain a plurality of control input sequences corresponding to the predicted state sequence; wherein the control input sequence comprises a plurality of control input data corresponding to a preset time step, and each of the control input data comprises an angular velocity and a heading angle of rotation that the cabin needs to take in the next period under an actual state; Performing optimization calculation according to the multiple control input sequences and a preset cost function, and determining the control input sequence that minimizes the cost function value as the target control input sequence; The cabin rotation angular velocity corresponding to the first control input data in the target control input sequence is determined as the target yaw angular velocity, and the cabin heading angle corresponding to the first control input data in the target control input sequence is determined as the target yaw angle.

8. An offshore wind turbine yaw control device based on MPC, characterized in that: It includes at least one control processor and a memory for communicating with the at least one control processor; the memory stores instructions that can be executed by the at least one control processor, and the instructions are executed by the at least one control processor so that the at least one control processor can execute the MPC-based offshore wind turbine yaw control method as described in any one of claims 1 to 7.

9. An electronic device, characterized in that: It comprises the MPC-based offshore wind turbine yaw control device as claimed in claim 8.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores computer-executable instructions, and the computer-executable instructions are used to enable a computer to execute the MPC-based offshore wind turbine yaw control method according to any one of claims 1 to 7.

Images