Wind speed acquisition method, control device and storage medium of wind turbine generator system

Abstract

The present application relates to the technical field of wind power, and particularly provides a wind turbine incoming flow wind speed acquisition method, a control device and a storage medium, aiming to solve the problem of how to more accurately acquire the incoming flow wind speed of a wind turbine. To this end, the present application acquires monitoring control system data and terrain parameters of the wind turbine, and according to the monitoring control system data and the terrain parameters, a trained time convolution network model is applied to acquire the incoming flow wind speed of the wind turbine. The influence of the monitoring control system data and the terrain parameters on the incoming flow wind speed is comprehensively considered, the relationship between the monitoring control system data and the incoming flow wind speed is determined, the terrain parameters of the wind turbine are considered, and the influence of the terrain factors on the incoming flow wind speed is reduced. The time convolution network model is introduced, the law and uncertainty of the wind speed in time sequence are considered, the robustness and generalization of the process of acquiring the incoming flow wind speed of the wind turbine are effectively increased, and a more accurate incoming flow wind speed is acquired.

Description

Technical Field

[0001] This invention relates to the field of wind power technology, specifically providing a method, control device, and storage medium for obtaining the incoming wind speed of a wind turbine. Background Technology

[0002] With the rapid development of wind power technology and its gradual large-scale grid connection, wind power is playing an increasingly important role in the electricity market. Although the leapfrog development of wind power has alleviated some of the energy crisis, the drastic fluctuations in the output power, frequency, and phase of wind farms have seriously affected the stability of the power system. Wind power prediction technology is one of the necessary ways to mitigate the adverse effects of wind power grid connection on the power system and increase the proportion of wind power grid connection. This requires accurately grasping the relationship between wind speed and power in wind farms and establishing a wind farm power characteristic model. The power characteristics of wind turbine units are an important indicator for evaluating their performance, which directly affects the calculation of power generation. The key to evaluation lies in how to accurately obtain the incoming wind speed at the nacelle. Due to the multiple influences of atmospheric motion, topographic factors, and the interaction between the wind turbine and the nacelle with the incoming wind conditions, the wind speed measured by the nacelle anemometer cannot truly reflect the incoming wind conditions at the wind turbine hub, i.e., the free-flowing wind speed. Since the nacelle anemometer is usually installed at the rear of the nacelle behind the wind turbine, the free flow is greatly disturbed when the wind turbine rotates and passes through the blades; and the free flow is also affected when it passes through the nacelle. Therefore, the changes in the free flow caused by the wind turbine's operating state and its own structure are the main sources of error for the nacelle anemometer.

[0003] Currently, wind turbine nacelle wind speed correction is mainly achieved through two methods: (1) theoretical calculation correction method, which directly uses the monitoring data and performance parameters of the turbine to correct the nacelle wind speed according to the wind power operation principle. In engineering applications, it is greatly affected by complex operating conditions and is highly dependent on the amount of measured data; (2) CFD (Computational Fluid Dynamics) based nacelle wind speed correction method, which uses different turbulence models to numerically simulate the wind turbine and uses different azimuth angle models to simulate the rotor rotation period, and finally obtains the velocity field above the nacelle; (3) function fitting method, which supplements the observation of the incoming wind speed in front of the wind turbine and directly fits the nacelle transfer function (Nacelle Transfer) between the observed wind speed and the nacelle wind speed. The IEC standard (61400-12-2:2013) specifies the specific methods and requirements for NTF (Network Function). This method is simple and easy to use and is widely used in practice. However, this method is highly dependent on the quality of the observation data and requires the removal of sectors affected by obstacles in the observation and nacelle wind speed data according to the wind direction. It is difficult to express the actual operating environment of the wind turbine and is greatly affected by the terrain.

[0004] Accordingly, there is a need in the field for a new method for obtaining the incoming wind speed of wind turbines to solve the above problems. Summary of the Invention

[0005] To overcome the above-mentioned shortcomings, the present invention is proposed to provide a solution, or at least a partial solution, to the problem of how to more accurately obtain the incoming wind speed of a wind turbine.

[0006] In a first aspect, the present invention provides a method for obtaining the incoming wind speed of a wind turbine generator, the method comprising:

[0007] Acquire the monitoring and control system data and terrain parameters of the wind turbine;

[0008] Based on the monitoring and control system data and the terrain parameters, a pre-trained temporal convolutional network model is applied to obtain the incoming wind speed of the wind turbine.

[0009] In one technical solution of the above-mentioned method for obtaining the incoming wind speed of a wind turbine, the step of "obtaining the incoming wind speed of the wind turbine by applying a pre-trained temporal convolutional network model based on the monitoring and control system data and the terrain parameters" includes:

[0010] Model input data is obtained based on the monitoring and control system data and the terrain parameters, and the model input data is normalized.

[0011] The normalized model input data is input into the trained temporal convolutional network model to obtain the incoming wind speed of the wind turbine.

[0012] In one technical solution of the above-mentioned method for obtaining the incoming wind speed of a wind turbine, the step of "inputting the normalized model input data into the trained temporal convolutional network model to obtain the incoming wind speed of the wind turbine" includes:

[0013] The normalized model input data is input into the trained temporal convolutional network model to obtain the output data of the temporal convolutional network model;

[0014] The output data is denormalized to obtain the incoming wind speed of the wind turbine.

[0015] In one technical solution of the above-mentioned method for obtaining the incoming wind speed of a wind turbine, the method further includes training the temporal convolutional network model according to the following steps:

[0016] A model sample set for constructing the temporal convolutional network model is provided, wherein the model sample set includes the monitoring and control system data and the terrain parameters.

[0017] The temporal convolutional network model is trained based on the model sample set to obtain a trained temporal convolutional network model.

[0018] In one technical solution of the above-mentioned method for obtaining the incoming wind speed of a wind turbine, the model sample set includes a model training set, and the step of "training the temporal convolutional network model based on the model sample set" includes:

[0019] Obtain the wind speed from the wind turbine's lidar.

[0020] The training labels for the model training are obtained based on the wind speed measured by the wind-measuring lidar.

[0021] The monitoring and control system data and the terrain parameters are obtained based on the model training set.

[0022] Model input data is obtained based on the monitoring and control system data and the terrain parameters, and the model input data is normalized.

[0023] The temporal convolutional network model is iteratively trained based on the normalized model input data and the training labels.

[0024] In one technical solution of the above-mentioned method for obtaining the incoming wind speed of a wind turbine, the step of "training the temporal convolutional network model based on the model sample set" further includes:

[0025] The trained temporal convolutional network model is validated using the model test set.

[0026] When the model passes validation, the temporal convolutional network model is used as the trained temporal convolutional network model; and / or,

[0027] The monitoring and control system data includes the monitored wind direction of the wind turbine. The step of "obtaining the training labels for the model training based on the wind speed measured by the wind-measuring lidar" includes:

[0028] Based on the monitored wind direction, wind speeds measured by lidar in sectors affected by obstacles are excluded.

[0029] The wind speeds from the removed wind-measuring lidar were normalized.

[0030] The normalized wind speed measured by the lidar is used as the training label for the model training.

[0031] In one technical solution of the above-mentioned method for obtaining the incoming wind speed of a wind turbine, the monitoring and control system data includes the monitored wind direction of the wind turbine, the terrain parameters include a terrain steepness index for each monitored wind direction, and the method further includes obtaining the terrain steepness index according to the following steps:

[0032] Based on the monitored wind direction, data from the monitoring and control system located in sectors affected by obstacles are excluded.

[0033] For each monitored wind direction in the data of the monitoring and control system after removal, multiple elevation points are obtained on the terrain profile centered on the wind turbine, and the absolute height deviation between any two adjacent elevation points is obtained.

[0034] The terrain steepness index for each monitored wind direction is obtained based on the absolute height deviation, the hub height of the wind turbine, and the diameter of the wind turbine.

[0035] In one technical solution of the above-mentioned method for obtaining the incoming wind speed of a wind turbine, the terrain parameters include the terrain inclination angle of each terrain profile, and the method further includes obtaining the terrain inclination angle according to the following formula:

[0036]

[0037] Where θ is the terrain slope angle; P is the number of absolute height deviations; c is the distance between two adjacent elevation points; z i For the i-th absolute height deviation; and / or,

[0038] The monitoring and control system data includes the nacelle wind speed of the wind turbine. The step of "obtaining model input data based on the monitoring and control system data and the terrain parameters" includes:

[0039] The removed cabin wind speed and the monitored wind direction, as well as the steepness index and terrain inclination corresponding to each removed monitored wind direction, are used as the model input data.

[0040] In a second aspect, a control device is provided, comprising a processor and a storage device, the storage device being adapted to store a plurality of program codes, the program codes being adapted to be loaded and run by the processor to execute the wind turbine incoming wind speed acquisition method described in any of the above-described technical solutions.

[0041] In a third aspect, a computer-readable storage medium is provided, wherein a plurality of program codes are stored therein, the program codes being adapted to be loaded and run by a processor to perform the wind turbine wind speed acquisition method described in any of the above-described technical solutions.

[0042] The present invention comprises one or more of the following technical solutions:

[0043] Beneficial effects:

[0044] In implementing the technical solution of this invention, the invention acquires monitoring and control system data and terrain parameters of the wind turbine generator. Based on the monitoring and control system data and terrain parameters, a trained temporal convolutional network model is applied to obtain the incoming wind speed of the wind turbine generator. Through the above configuration, this invention obtains the incoming wind speed based on monitoring and control system data and terrain parameters, comprehensively considering the influence of monitoring and control system data and terrain parameters on the incoming wind speed. While determining the intrinsic relationship between the monitoring and control system data and the incoming wind speed, it also considers the terrain parameters of the wind turbine generator, which can reduce the influence of terrain factors on the incoming wind speed. Simultaneously, introducing a temporal convolutional network model to obtain the incoming wind speed can consider the temporal regularity and uncertainty of wind speed, thereby effectively increasing the robustness and generalization of the process of obtaining the incoming wind speed of the wind turbine generator, resulting in a more accurate acquisition of the incoming wind speed. Attached Figure Description

[0045] The disclosure of this invention will become more readily understood with reference to the accompanying drawings. It will be readily understood by those skilled in the art that these drawings are for illustrative purposes only and are not intended to limit the scope of protection of this invention. Wherein:

[0046] Figure 1 This is a schematic flowchart of the main steps of a method for obtaining the incoming wind speed of a wind turbine according to an embodiment of the present invention;

[0047] Figure 2 This is a schematic flowchart of the main steps of a method for obtaining the incoming wind speed of a wind turbine according to an embodiment of the present invention.

[0048] Figure 3 This is a schematic diagram of a monitoring and control system for removing data from sectors affected by obstacles, according to one embodiment of the present invention.

[0049] Figure 4 This is a schematic diagram of the main model structure of a temporal convolutional network model according to one embodiment of the present invention;

[0050] Figure 5 This is a scatter plot comparing the cabin wind speed and the incoming airflow speed obtained according to an embodiment of the present invention.

[0051] Figure 6 This is a time-series comparison diagram of cabin wind speed, incoming air speed obtained according to an embodiment of the present invention, and actual incoming air speed. Detailed Implementation

[0052] Some embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0053] In the description of this invention, "module" and "processor" can include hardware, software, or a combination of both. A module can include hardware circuitry, various suitable sensors, communication ports, memory, and may also include software components, such as program code, or a combination of software and hardware. A processor can be a central processing unit, microprocessor, image processor, digital signal processor, or any other suitable processor. The processor has data and / or signal processing capabilities. The processor can be implemented in software, in hardware, or a combination of both. Non-transitory computer-readable storage media includes any suitable medium capable of storing program code, such as magnetic disks, hard disks, optical disks, flash memory, read-only memory, random access memory, etc. The term "A and / or B" means all possible combinations of A and B, such as only A, only B, or A and B. The terms "at least one A or B" or "at least one of A and B" have a similar meaning to "A and / or B" and can include only A, only B, or A and B. The singular terms "a" or "this" can also include plural forms.

[0054] Here we will first explain some of the terms involved in this invention.

[0055] Nacelle wind speed refers to the wind speed measured by an anemometer (such as a wind-measuring lidar) in the nacelle behind the wind turbine rotor.

[0056] Incoming wind speed refers to the wind speed in front of the wind turbine rotor.

[0057] Due to the influence of the wind turbine and nacelle, there will be a difference between the nacelle wind speed and the incoming wind speed.

[0058] Current traditional methods for correcting nacelle wind speeds mainly have two problems:

[0059] (1) The above methods all take into account the influence factors of the wind turbine itself. However, in complex terrain, the turbulence and tilt angle of the incoming wind are different under different wind directions. After passing through the wind turbine and nacelle, they will exhibit different degrees of distortion. Therefore, the robustness of the model is poor under different working conditions.

[0060] (2) Wind speed is a time-series data, and there is a strong regularity between time series. When establishing the nacelle transfer function, the above methods either use physical models to back-calculate the incoming wind speed or use statistical models to study the distribution relationship between the incoming wind speed and the nacelle wind speed. They do not consider the correlation between wind speed time series and ignore the dynamic changes of the observed wind speed and the nacelle wind speed over time. When the wind speed changes drastically, the generalization ability of the model is poor.

[0061] Accordingly, there is a need in the field for a new method for obtaining the incoming wind speed of wind turbines to solve the above problems.

[0062] See appendix Figure 1 , Figure 1 This is a schematic flowchart illustrating the main steps of a method for obtaining the incoming wind speed of a wind turbine according to an embodiment of the present invention. Figure 1 As shown, the method for obtaining the incoming wind speed of the wind turbine in this embodiment of the invention mainly includes the following steps S101-S102.

[0063] Step S101: Obtain monitoring and control system data and terrain parameters of the wind turbine.

[0064] In this embodiment, monitoring and control system data and terrain parameters of the wind turbine can be acquired. The monitoring and control system (SCADA) is a system used to measure data from the wind turbine.

[0065] In one implementation, the monitoring and control system data may include the nacelle wind speed and monitored wind direction of the wind turbine.

[0066] In one implementation, the terrain parameters may include the terrain steepness index and terrain inclination angle of the wind turbine. The terrain steepness index refers to the percentage by which the steepness of the terrain profile where the wind turbine is located exceeds a certain limit. The terrain inclination angle refers to the angle between the terrain profile where the wind turbine is located and the actual horizontal plane. A terrain profile is a vertical cross-section along a straight line on the earth's surface, showing the undulations of the terrain along the profile line.

[0067] Step S102: Based on the monitoring and control system data and terrain parameters, apply a pre-trained temporal convolutional network model to obtain the incoming wind speed of the wind turbine.

[0068] In this embodiment, the incoming wind speed of the wind turbine can be obtained by applying a temporal convolutional network model based on monitoring and control system data and terrain parameters. The temporal convolutional network (TCN) is an algorithm that can be used to solve time series forecasting problems.

[0069] In one implementation, the nacelle wind speed of the wind turbine, the monitored wind direction, the terrain steepness index, and the terrain inclination can be input into a trained temporal convolutional network model to predict the incoming wind speed of the wind turbine and obtain the incoming wind speed of the wind turbine.

[0070] Based on steps S101-S102 above, this embodiment of the invention acquires monitoring and control system data and terrain parameters of the wind turbine generator. Using this data and these parameters, a trained temporal convolutional network model is applied to obtain the incoming wind speed of the wind turbine generator. Through this configuration, this embodiment of the invention acquires the incoming wind speed based on monitoring and control system data and terrain parameters, comprehensively considering the influence of both data and parameters on the incoming wind speed. In addition to determining the intrinsic relationship between the monitoring and control system data and the incoming wind speed, it also considers the terrain parameters of the wind turbine generator, thus reducing the impact of terrain factors on the incoming wind speed. Furthermore, introducing a temporal convolutional network model to acquire the incoming wind speed allows consideration of the temporal regularity and uncertainty of wind speed, thereby effectively increasing the robustness and generalization of the process of acquiring the incoming wind speed of the wind turbine generator, resulting in a more accurate acquisition of the incoming wind speed.

[0071] The following is a further explanation of step S102.

[0072] In one embodiment of the present invention, step S102 may further include steps S1021 and S1022:

[0073] Step S1021: Obtain model input data based on monitoring and control system data and terrain parameters, and normalize the model input data.

[0074] In this embodiment, model input data can be obtained based on monitoring and control system data and terrain parameters, and the model input data can be normalized.

[0075] In one implementation, the monitoring and control system data located in the obstacle sector can be removed based on the monitored wind direction in the monitoring and control system data; and the cabin wind speed and monitored wind direction in the removed monitoring and control data, as well as the steepness index and terrain inclination corresponding to each removed monitored wind direction, can be used as model input data.

[0076] Specifically, there should be no large obstacles (such as buildings, trees, other wind turbines, etc.) within the wind turbine and wind measurement equipment sector. Obstacles can be considered as cylinders with a diameter equal to the base diameter of the tower and a height equal to the top height of the tower. The equivalent rotor diameter of the obstacle can be determined using the following formula (1):

[0077]

[0078] Among them, De The equivalent rotor diameter of the obstacle; l n The height of the obstacle; l w The height of an obstacle as seen by a wind turbine or wind measuring equipment.

[0079] Please refer to the appendix. Figure 3 , Figure 3 This is a schematic diagram illustrating a monitoring and control system for removing data from sectors affected by obstacles, according to one embodiment of the present invention. Figure 3 The x-coordinate is the relative distance, that is, D. e / L e The vertical axis represents the sector angle affected by the obstacle. For example... Figure 3 As shown, monitoring and control system data located in the following sectors can be excluded:

[0080] 1) The wind measurement equipment (wind measurement tower or remote sensing equipment) is in the wake of the wind turbine being measured;

[0081] 2) The wind measuring equipment is located in the wake of a nearby operating wind turbine;

[0082] 3) The wind turbine under test is in the wake of a nearby operating wind turbine;

[0083] 4) The wind measuring equipment is in the wake of a large obstacle;

[0084] 5) The wind turbine under test is in the wake of a large obstacle;

[0085] 6) Combinations of all the above situations.

[0086] The sector angle affected by the obstacle can be determined using the following formula (2):

[0087] α = 1.3arctan(2.5D) e / L e +0.15)+10 (2)

[0088] Where α is the sector angle, L e This refers to the actual horizontal distance between the obstacle and the center of the target wind turbine or the location of the wind measurement equipment.

[0089] In one implementation, the model input data can be normalized according to the following formula (3).

[0090]

[0091] Where, x * Here, x represents the normalized model input data. min x is the minimum value of the input data to the model. max The maximum value of the input data for the model.

[0092] Step S1022: Input the normalized model input data into the trained temporal convolutional network model to obtain the incoming wind speed of the wind turbine.

[0093] In this embodiment, step S1022 may further include steps S10221 and S10222:

[0094] Step S10221: Input the normalized model input data into the trained temporal convolutional network model and obtain the output data of the temporal convolutional network model.

[0095] Step S10222: Perform inverse normalization on the output data to obtain the incoming wind speed of the wind turbine.

[0096] In this embodiment, normalized model input data can be input into a trained temporal convolutional network model to obtain the output data of the temporal convolutional network model. The output data is then denormalized to obtain the incoming wind speed of the wind turbine.

[0097] In one embodiment of the present invention, the present invention can train a temporal convolutional network model according to the following steps S103 and S104:

[0098] Step S103: Construct a model sample set for the temporal convolutional network model, wherein the model sample set includes monitoring and control system data and terrain parameters.

[0099] In one implementation, the model sample set may include a model training set and a model test set.

[0100] In one implementation, the model sample set can be divided into a model training set and a model test set at a ratio of 80% and 20%, respectively.

[0101] Step S104: Train the temporal convolutional network model based on the model sample set to obtain a trained temporal convolutional network model.

[0102] In one embodiment, step S104 may further include steps S1041 to S1045:

[0103] Step S1041: Obtain the wind speed from the wind turbine's lidar.

[0104] In this embodiment, the wind speed at the hub height of the wind turbine can be obtained using a wind-measuring lidar as the wind speed measured by the lidar.

[0105] Step S1042: Obtain training labels for model training based on wind speed measured by the wind-measuring lidar.

[0106] In this embodiment, step S1042 may further include steps S10421 to S10423:

[0107] Step S10421: Based on the monitored wind direction, eliminate the wind speed measured by lidar located in the sector affected by obstacles.

[0108] Step S10422: Normalize the wind speed from the removed wind lidar.

[0109] Step S10423: Use the normalized wind speed from the wind-measuring lidar as the training label for model training.

[0110] Step S1043: Obtain monitoring and control system data and terrain parameters based on the model training set.

[0111] In this embodiment, monitoring and control system data and terrain parameters can be obtained from the model training set.

[0112] Step S1044: Obtain model input data based on monitoring and control system data and terrain parameters, and normalize the model input data.

[0113] In this embodiment, the cabin wind speed and monitored wind direction after removal, as well as the steepness index and terrain inclination corresponding to each removed monitored wind direction, are used as model input data.

[0114] Step S1045: Based on the normalized model input data and training labels, perform iterative training on the temporal convolutional network model.

[0115] In this embodiment, normalized model input data and training labels can be used to perform iterative training on the temporal convolutional network model.

[0116] In one implementation, the terrain steepness index for each monitored wind direction can be obtained according to the following steps S201 to S203:

[0117] Step S201: Based on the monitored wind direction, exclude monitoring and control system data located in sectors affected by obstacles.

[0118] In this embodiment, monitoring and control system data located in the sector affected by obstacles under each monitored wind direction can be removed.

[0119] Step S202: For each monitored wind direction in the data of the monitoring and control system after removal, obtain multiple elevation points on the terrain profile centered on the wind turbine, and obtain the absolute height deviation between every two adjacent elevation points.

[0120] In this embodiment, for each monitored wind direction, multiple elevation points are determined on the terrain profile centered on the wind turbine (e.g., one elevation point is determined every 30m), and the absolute height deviation between two adjacent elevation points is calculated.

[0121] In one implementation, the absolute height deviation between two adjacent elevation points can be calculated according to the following formula (4):

[0122] Δz i =|(elevation) i -elevation i-1 (4)

[0123] Where, Δz i For the i-th absolute height deviation, elevation i For the i-th elevation point, elevation i-1 Let be the (i-1)th elevation point.

[0124] Step S203: Based on the absolute height deviation, the hub height of the wind turbine, and the diameter of the wind turbine, obtain the terrain steepness index for each monitored wind direction.

[0125] In this embodiment, the terrain steepness index of the wind turbine can be obtained under each monitored wind direction based on the absolute height deviation, the hub height of the wind turbine, and the diameter of the wind turbine.

[0126] In one implementation, based on the absolute height deviation obtained in step S202, the terrain steepness index for each terrain profile may include RIX. 0.04 RIX 0.06 RIX 0.08 It can be obtained by dividing the number of absolute height deviations of the terrain profile exceeding 0.04(D+H), 0.06(D+H), and 0.08(D+H) by the percentage of all absolute height deviations P. Where D is the rotor diameter of the wind turbine and H is the hub height of the wind turbine.

[0127] In one implementation, a digital topographic map with a radius of 20D centered on the wind turbine can be obtained using a digital elevation model (DEM) of the wind farm. This digital topographic map is used to determine the topographic profile centered on the wind turbine and multiple elevation points.

[0128] In one implementation, the topographic dip angle of each topographic profile can be obtained according to the following formula (5):

[0129]

[0130] Where θ is the terrain slope angle; P is the number of absolute height deviations; c is the distance between two adjacent elevation points; z i Let be the absolute height deviation of the i-th height.

[0131] In one embodiment, step S104 may further include steps S1046 and S1047:

[0132] Step S1046: Apply the model test set to validate the trained temporal convolutional network model.

[0133] Step S1047: When the model passes validation, use the temporal convolutional network model as the trained temporal convolutional network model.

[0134] In one implementation, see Appendix Figure 4 , Figure 4 This is a schematic diagram of the main model structure of a temporal convolutional network model according to one embodiment of the present invention. Figure 4 As shown, a temporal convolutional network model can be constructed. The input feature matrix of the temporal convolutional network model is X = [x1, x2, ..., x]. t ], where t is the time step. The dilated causal convolutional layer can be calculated according to the following formula (6):

[0135]

[0136] The activation function ReLU can be calculated according to the following formula (7):

[0137] ReLU(x) = max(0, x) (7)

[0138] Dropout can be calculated using the following formula (8):

[0139]

[0140] The residual connection can be calculated according to the following formula (9):

[0141] o=σ(x+F(x)) (9)

[0142] Where d represents the hole factor of the causal dilated convolution, k represents the size of the convolution kernel, rand(·) represents a random function, p represents the probability of inactivation, and σ represents the activation function.

[0143] The results obtained from the residual connection calculation are subjected to a one-dimensional full convolution. Based on the result of the one-dimensional full convolution and the result of the second Dropout calculation, the output matrix of the temporal convolutional network model can be obtained as Y = [y1, y2, ..., y]. t ].

[0144] In one implementation, the input, output, and hyperparameters of the temporal convolutional network model can be set. The parameters of the input layer are the normalized nacelle wind speed, monitored wind direction, terrain steepness index, and terrain inclination angle. The parameters of the output layer are the normalized incoming wind speed of the wind turbine. The model sample set can be divided into a training set and a test set, representing 80% and 20% respectively. The initial settings are: loss function MSE (Mean Square Error), optimizer Adam, learning rate 0.001, time step 12, dilation factor 2, 4 convolutional layers, dropout 0.2, and 200 training iterations.

[0145] The samples in the model training set are input into the temporal convolutional network model for training. The temporal convolutional network model is trained and adjusted by changing the learning rate, number of iterations, and model parameters until the optimal model parameters are obtained.

[0146] In one implementation, RMSE (Root mean squared error) is used as the model evaluation criterion. Samples from the model test set are input into the optimal temporal convolutional network model, and the effectiveness of the temporal convolutional network model is verified according to the following formula (10):

[0147]

[0148] Where n is the number of calculated values ​​output by the model. y is the calculated value of the i-th incoming wind speed output by the model. i Let be the wind speed of the i-th wind-measuring lidar.

[0149] In one implementation, a root mean square error threshold can be set. When the calculated root mean square error is less than the root mean square error threshold, the model is considered valid, that is, the valid model can be used as a trained temporal convolutional network model.

[0150] In one implementation, the nacelle wind speed, monitored wind direction, terrain steepness index, and terrain inclination of the target wind turbine at any given time can be used as model input data. After normalization, the data is input into a trained temporal convolutional network model, and the model output data is inversely normalized to obtain the incoming wind speed of the target wind turbine.

[0151] In one implementation, see Appendix Figure 2 , Figure 2 This is a schematic flowchart illustrating the main steps of a method for obtaining the incoming wind speed of a wind turbine according to an embodiment of the present invention. Figure 2 As shown, the method for obtaining the incoming wind speed of a wind turbine may include the following steps S301 to S312:

[0152] Step S301: Data acquisition.

[0153] In this embodiment, the required data can be collected.

[0154] Step S302: Obtain the wind speed from the wind turbine's lidar.

[0155] In this embodiment, the wind speed of the wind turbine can be obtained based on the collected data using a lidar system.

[0156] Step S303: Obtain monitoring and control system data for the wind turbine.

[0157] In this embodiment, monitoring and control system data of the wind turbine can be obtained based on the collected data.

[0158] Step S304: Obtain digital high-rise map (DEM) data of the wind farm.

[0159] In this embodiment, digital high-rise map (DEM) data of the wind farm can be obtained based on the collected data.

[0160] Step S305: Remove data from sectors affected by obstacles.

[0161] In this embodiment, data located in sectors affected by obstacles can be removed.

[0162] Step S306: Calculate the terrain parameters of the wind turbine.

[0163] In this embodiment, the terrain parameters of the wind turbine can be calculated.

[0164] Step S307: Normalize the model input data of the wind turbine.

[0165] In this embodiment, the model input data can be normalized to obtain the model's input feature matrix.

[0166] Step S308: Set the parameters of the temporal convolutional network model.

[0167] In this embodiment, the parameters of the temporal convolutional network model can be set.

[0168] Step S309: Train the temporal convolutional network model based on the input feature matrix.

[0169] In this embodiment, the temporal convolutional network model can be trained based on the input feature matrix.

[0170] Step S310: Determine whether the training has terminated; if yes, proceed to step S311; if no, proceed to step S309.

[0171] In this implementation, it can be determined whether the training is complete.

[0172] Step S311: Obtain a temporal convolutional network model with optimal parameters.

[0173] In this embodiment, a temporal convolutional network model with optimal parameters can be obtained.

[0174] Step S312: Test the temporal convolutional network model.

[0175] In this embodiment, a temporal convolutional network model can be used for model testing.

[0176] Step S313: Verify the effectiveness of the temporal convolutional network model.

[0177] In this embodiment, the effectiveness of the temporal convolutional network model can be verified.

[0178] Step S314: Apply the validated model to obtain the incoming wind speed of the wind turbine.

[0179] In this embodiment, a validated (trained) model can be used to obtain the incoming wind speed of the wind turbine.

[0180] Please refer to the appendix. Figure 5 and attached Figure 6 , Figure 5 This is a scatter plot comparing the cabin wind speed and the incoming airflow speed obtained according to an embodiment of the present invention. Figure 6 This is a time-series comparison diagram of nacelle wind speed, incoming airflow speed obtained according to an embodiment of the present invention, and actual incoming airflow speed. Wherein, Figure 5 The horizontal axis represents the nacelle wind speed, and the vertical axis represents the incoming airflow speed. From Figure 5 As can be seen from the data, with y=x as the reference, the incoming wind speed of the wind turbine obtained through the embodiments of the present invention (TCN wind speed) is more concentrated than the nacelle wind speed. Figure 6 The horizontal axis represents time, and the vertical axis represents wind speed. From Figure 6 As can be seen from the above, the incoming wind speed of the wind turbine obtained through the embodiments of the present invention (TCN wind speed) is closer to the actual incoming wind speed than the nacelle wind speed.

[0181] It should be noted that although the steps in the above embodiments are described in a specific order, those skilled in the art will understand that in order to achieve the effects of the present invention, different steps do not necessarily have to be executed in such an order. They can be executed simultaneously (in parallel) or in other orders, and these variations are all within the scope of protection of the present invention.

[0182] Those skilled in the art will understand that all or part of the processes in the method of the above embodiment of the present invention can also be implemented by a computer program instructing related hardware. The computer program can be stored in a computer-readable storage medium, and when executed by a processor, it can implement the steps of the various method embodiments described above. The computer program includes computer program code, which can be in the form of source code, object code, executable file, or some intermediate form. The computer-readable storage medium can include any entity or device capable of carrying the computer program code, a medium, a USB flash drive, a portable hard drive, a magnetic disk, an optical disk, a computer memory, a read-only memory, a random access memory, an electrical carrier signal, a telecommunication signal, and a software distribution medium, etc. It should be noted that the content included in the computer-readable storage medium can be appropriately added or removed according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, the computer-readable storage medium does not include electrical carrier signals and telecommunication signals.

[0183] Furthermore, the present invention also provides a control device. In one embodiment of the control device according to the present invention, the control device includes a processor and a storage device. The storage device can be configured to store a program for executing the wind turbine wind speed acquisition method of the above-described method embodiments. The processor can be configured to execute the program in the storage device, which includes, but is not limited to, a program for executing the wind turbine wind speed acquisition method of the above-described method embodiments. For ease of explanation, only the parts related to the embodiments of the present invention are shown. For specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. This control device can be a control device device comprising various electronic devices.

[0184] Furthermore, the present invention also provides a computer-readable storage medium. In one embodiment of the computer-readable storage medium according to the present invention, the computer-readable storage medium can be configured to store a program for executing the wind turbine incoming wind speed acquisition method of the above-described method embodiments. This program can be loaded and run by a processor to implement the wind turbine incoming wind speed acquisition method described above. For ease of explanation, only the parts related to the embodiments of the present invention are shown; for specific technical details not disclosed, please refer to the method section of the embodiments of the present invention. The computer-readable storage medium can be a storage device device comprising various electronic devices. Optionally, in the embodiments of the present invention, the computer-readable storage medium is a non-transitory computer-readable storage medium.

[0185] Furthermore, it should be understood that since the various modules are only provided to illustrate the functional units of the device of the present invention, the physical devices corresponding to these modules may be the processor itself, or a part of the processor's software, hardware, or a combination of software and hardware. Therefore, the number of modules shown in the figures is merely illustrative.

[0186] Those skilled in the art will understand that the various modules in the device can be adaptively split or combined. Such splitting or combining of specific modules will not cause the technical solution to deviate from the principles of the present invention; therefore, the technical solutions after splitting or combining will fall within the protection scope of the present invention.

[0187] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after such changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A method for obtaining the incoming wind speed of a wind turbine generator, characterized in that, The method includes: Acquire the monitoring and control system data and terrain parameters of the wind turbine; Based on the monitoring and control system data and the terrain parameters, a pre-trained temporal convolutional network model is applied to obtain the incoming wind speed of the wind turbine. The method further includes training the temporal convolutional network model according to the following steps: A model sample set for constructing the temporal convolutional network model is provided, wherein the model sample set includes the monitoring and control system data and the terrain parameters. The temporal convolutional network model is trained based on the model sample set to obtain a trained temporal convolutional network model. The model sample set includes a model training set, and the step of "training the temporal convolutional network model based on the model sample set" includes: Obtain the wind speed from the wind turbine's lidar. The training labels for the model training are obtained based on the wind speed measured by the wind-measuring lidar. The monitoring and control system data and the terrain parameters are obtained based on the model training set; Model input data is obtained based on the monitoring and control system data and the terrain parameters, and the model input data is normalized. Based on the normalized model input data and the training labels, the temporal convolutional network model is trained iteratively. The monitoring and control system data includes the monitored wind direction of the wind turbine, and the terrain parameters include a terrain steepness index for each monitored wind direction. The method further includes obtaining the terrain steepness index according to the following steps: Based on the monitored wind direction, data from the monitoring and control system located in sectors affected by obstacles are excluded. For each monitored wind direction in the data of the monitoring and control system after removal, multiple elevation points are obtained on the terrain profile centered on the wind turbine, and the absolute height deviation between any two adjacent elevation points is obtained. Based on the absolute height deviation, the hub height of the wind turbine, and the diameter of the wind turbine, the terrain steepness index for each monitored wind direction is obtained; The terrain parameters include the terrain dip angle of each terrain profile, and the method further includes obtaining the terrain dip angle according to the following formula: in, The terrain inclination angle; The number of absolute deviations in height; This represents the distance between two adjacent elevation points. For the i-th absolute height deviation; and / or, The monitoring and control system data includes the nacelle wind speed of the wind turbine. The step of "obtaining model input data based on the monitoring and control system data and the terrain parameters" includes: The removed cabin wind speed and the monitored wind direction, as well as the steepness index and terrain inclination corresponding to each removed monitored wind direction, are used as the model input data.

2. The method for obtaining the incoming wind speed of a wind turbine generator according to claim 1, characterized in that, The step of "obtaining the incoming wind speed of the wind turbine by applying a pre-trained temporal convolutional network model based on the monitoring and control system data and the terrain parameters" includes: Model input data is obtained based on the monitoring and control system data and the terrain parameters, and the model input data is normalized. The normalized model input data is input into the trained temporal convolutional network model to obtain the incoming wind speed of the wind turbine.

3. The method for obtaining the incoming wind speed of a wind turbine generator according to claim 2, characterized in that, The step of "inputting the normalized model input data into the trained temporal convolutional network model to obtain the incoming wind speed of the wind turbine" includes: The normalized model input data is input into the trained temporal convolutional network model to obtain the output data of the temporal convolutional network model; The output data is denormalized to obtain the incoming wind speed of the wind turbine.

4. The method for obtaining the incoming wind speed of a wind turbine generator according to claim 1, characterized in that, The model sample set includes a model test set, and the step of "training the temporal convolutional network model based on the model sample set" further includes: The trained temporal convolutional network model is validated using the model test set. When the model passes validation, the temporal convolutional network model is used as the trained temporal convolutional network model.

5. The method for obtaining the incoming wind speed of a wind turbine generator according to claim 1, characterized in that, The monitoring and control system data includes the monitored wind direction of the wind turbine. The step of "obtaining the training labels for the model training based on the wind speed measured by the wind-measuring lidar" includes: Based on the monitored wind direction, wind speeds measured by lidar in sectors affected by obstacles are excluded. The wind speeds from the removed wind-measuring lidar were normalized. The normalized wind speed measured by the lidar is used as the training label for the model training.

6. A control device, comprising a processor and a storage device, said storage device being adapted to store a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by the processor to perform the method for obtaining the incoming wind speed of the wind turbine unit as described in any one of claims 1 to 5.

7. A computer-readable storage medium storing a plurality of program codes, characterized in that, The program code is adapted to be loaded and run by a processor to perform the method for obtaining the incoming wind speed of the wind turbine unit as described in any one of claims 1 to 5.

Images