A control method and device of a wind turbine laser radar and a medium

Abstract

The application relates to the technical field of wind power, and discloses a control method and device of a wind turbine laser radar and a medium, which comprises the following steps: acquiring an initial estimated wind speed, filtering the acquired initial estimated wind speed through Kalman filtering, so as to correct the initial estimated wind speed and obtain an optimal estimated value. Then, each optimal estimated value is calculated to obtain an expected pitch angle increment, the expected pitch angle increment is corrected through a BP neural network to obtain a target pitch angle increment, and the target pitch angle increment is superimposed on a given pitch angle increment to output an actual rotating speed. In this way, the wind speed of the laser radar is continuously predicted and optimized through Kalman filtering, so that the laser radar can accurately predict the wind speed even if there is blade shielding and the like, and the wind turbine control effect is improved. In addition, the expected pitch angle increment is corrected through the BP neural network, real-time adjustment of the pitch angle increment is realized, the rotating speed of the wind wheel is prevented from being too high, and the loss of power generation is reduced.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of wind power, in particular to a wind turbine laser radar control method, device and medium. BACKGROUND

[0002] The laser radar is a remote sensing device for detecting by using laser, and the laser radar feedforward control technology is to optimize the fan control strategy by using the control type laser radar, so as to reduce the load, improve the applicability and power of the fan, and prolong the service life of the fan. Therefore, the laser radar feedforward control technology is crucial for the development of wind power.

[0003] At present, in the laser radar feedforward control, since the laser radar device is installed at the end of the cabin, the measured wind speed is affected by the rotation of the blade, so it cannot represent the real wind speed near the hub. That is, due to the influence of factors such as blade rotation, the estimated wind speed is inaccurate, thereby reducing the control effect of the fan.

[0004] In addition, in the laser radar feedforward control, since the variable gain PI algorithm adopts the speed feedback mode, the variable pitch cannot act in advance, and needs to be pitched when the wind blows, which will cause the change of the generator speed to lag behind the change of the wind speed, thereby causing the generator speed to fluctuate greatly, and the problem of overspeed of the wind wheel speed. When the generator speed reaches the limit, the wind turbine will stop due to overspeed, resulting in loss of power generation.

[0005] Therefore, how to solve the problem of inaccurate estimated wind speed caused by factors such as blade rotation, improve the estimated wind speed reaching the hub, improve the control effect of the fan, and at the same time, solve the problem of overspeed of the wind wheel speed, is a problem to be solved by those skilled in the art. SUMMARY

[0006] The purpose of the present application is to provide a wind turbine laser radar control method, device and medium, which solves the problem of inaccurate estimated wind speed caused by factors such as blade rotation, improves the estimated wind speed reaching the hub, improves the control effect of the fan, and at the same time, solves the problem of overspeed of the wind wheel speed.

[0007] To solve the above technical problems, the present application provides a wind turbine laser radar control method, comprising:

[0008] obtaining an initial estimated wind speed;

[0009] filtering the initial estimated wind speed by Kalman filtering to correct the initial estimated wind speed to obtain an optimal estimation value;

[0010] calculating each optimal estimation value to obtain an expected pitch angle increment;

[0011] The target pitch angle increment is obtained by correcting the expected pitch angle increment using a BP neural network.

[0012] The target pitch angle increment is superimposed on the given pitch angle increment to output the actual rotational speed.

[0013] Preferably, the step of filtering the initial estimated wind speed using Kalman filtering to correct the initial estimated wind speed and obtain the optimal estimate includes:

[0014] A lidar wind measurement model is constructed based on a pre-established reference coordinate system;

[0015] The lidar wind measurement model is transformed into a discretized state-space equation;

[0016] The optimal estimate is obtained by predicting the wind speed of the next state based on the discretized state-space equation and the initial estimated wind speed.

[0017] Preferably, after obtaining the optimal estimate, the method further includes:

[0018] The optimal estimate is optimized based on the positive definite matrix.

[0019] Preferably, the step of correcting the expected pitch angle increment using a BP neural network to obtain the target pitch angle increment includes:

[0020] Neural network training samples are constructed using a pre-defined reconstruction method;

[0021] The hidden layers and output layer of the BP neural network are calculated;

[0022] The weights of the BP neural network are adjusted using the error gradient descent method, and the thresholds of the hidden layer and the output layer are updated.

[0023] After constructing the Jacobian matrix, the target pitch angle increment is calculated based on the Jacobian matrix.

[0024] Preferably, the method further includes, before calculating the hidden layers and output layers in the BP neural network:

[0025] The input layer, hidden layer, and output layer of the BP neural network are initialized respectively.

[0026] Preferably, the preset reconstruction method is a phase space reconstruction method.

[0027] Preferably, the lidar is a continuous wave lidar.

[0028] To address the aforementioned technical problems, this application also provides a control device for a wind turbine lidar, comprising:

[0029] an acquisition module configured to acquire an initial estimated wind speed;

[0030] a filtering module configured to filter the initial estimated wind speed by Kalman filtering to correct the initial estimated wind speed to obtain an optimal estimated value;

[0031] a calculation module configured to calculate each of the optimal estimated values to obtain an expected pitch angle increment;

[0032] a correction module configured to correct the expected pitch angle increment by a BP neural network to obtain a target pitch angle increment;

[0033] a superposition module configured to superimpose the target pitch angle increment on a given pitch angle increment to output an actual rotating speed.

[0034] To solve the above technical problem, the application further provides a control device of a wind turbine laser radar, comprising a memory configured to store a computer program;

[0035] a processor configured to execute the computer program to implement the steps of the control method of the wind turbine laser radar.

[0036] To solve the above technical problem, the application further provides a computer readable storage medium, wherein the computer readable storage medium stores a computer program, and the computer program is executed by a processor to implement the steps of the control method of the wind turbine laser radar.

[0037] The control method of the wind turbine laser radar provided by the application comprises the following steps: acquiring an initial estimated wind speed, filtering the acquired initial estimated wind speed by Kalman filtering, correcting the initial estimated wind speed to obtain an optimal estimated value, calculating each of the optimal estimated values to obtain an expected pitch angle increment, correcting the expected pitch angle increment by a BP neural network to obtain a target pitch angle increment, and superimposing the target pitch angle increment on a given pitch angle increment to output an actual rotating speed. It can be seen that, by means of the technical solution provided by the application, the wind speed of the laser radar is continuously predicted and optimized by Kalman filtering, so that the wind speed can be accurately predicted even if the laser radar is affected by blade shielding and the like, and the control effect of the wind turbine is improved. In addition, the expected pitch angle increment is corrected by a BP neural network, so that the pitch angle increment is adjusted in real time, that is, the control of the variable pitch is realized, the rotating speed of the wind wheel is prevented from being too high, and the loss of power generation is reduced.

[0038] In addition, the application further provides a control device and a medium of a wind turbine laser radar, which correspond to the control method of the wind turbine laser radar described above and have the same effects. BRIEF DESCRIPTION OF DRAWINGS

[0039] In order to more clearly illustrate the embodiments of the present application, the drawings required to be used in the embodiments will be briefly introduced as follows. Obviously, the drawings described in the following embodiments are only some of the embodiments of the present application, and all other drawings obtained by those skilled in the art without creative labor based on these drawings also belong to the protection scope of the present application.

[0040] Figure 1 A flow chart of a wind turbine laser radar control method provided by an embodiment of the present application;

[0041] Figure 2 A schematic diagram of another wind turbine laser radar control method provided by an embodiment of the present application;

[0042] Figure 3 A schematic diagram of a reference coordinate system provided by an embodiment of the present application;

[0043] Figure 4 A flow chart of a wind turbine laser radar control method provided by another embodiment of the present application;

[0044] Figure 5 A flow chart of a wind turbine laser radar control method provided by another embodiment of the present application;

[0045] Figure 6 A schematic diagram of a BP neural network provided by an embodiment of the present application;

[0046] Figure 7 A structural diagram of a wind turbine laser radar control device provided by an embodiment of the present application;

[0047] Figure 8 A structural diagram of a wind turbine laser radar control device provided by another embodiment of the present application. DETAILED DESCRIPTION

[0048] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, but not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor also belong to the protection scope of the present application.

[0049] The core of the present application is to provide a wind turbine laser radar control method, device and medium, to ensure that the laser radar is not blocked by the blades, to improve the wind speed prediction accuracy, to solve the problem of wind wheel overspeed, and to reduce the loss of power generation.

[0050] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0051] LiDAR is a remote sensing device that uses lasers for detection. LiDAR feedforward control technology optimizes wind turbine control strategies using controlled LiDAR to reduce load, improve turbine applicability and power efficiency, and extend turbine lifespan. Therefore, LiDAR feedforward control technology is crucial for wind power development.

[0052] Currently, in lidar feedforward control, because the lidar device is installed at the end of the nacelle, the measured wind speed is affected by the blade rotation and therefore cannot represent the true wind speed near the hub. In other words, the predicted wind speed is inaccurate due to factors such as blade rotation, thus reducing the control effect of the wind turbine.

[0053] Furthermore, in lidar feedforward control, since the variable gain PI algorithm uses speed feedback, the pitch cannot be adjusted in advance and can only be adjusted when the wind blows. This causes the change in generator speed to lag behind the change in wind speed, resulting in large fluctuations in generator speed and the problem of wind turbine overspeed. When the generator speed reaches its limit, the wind turbine will shut down due to overspeed, resulting in a loss of power generation.

[0054] To address the aforementioned technical problems, including inaccurate wind speed predictions caused by blade rotation and other factors, improve the estimated wind speed reaching the hub, and enhance wind turbine control, this application provides a control method for a wind turbine lidar system. This method employs Kalman filtering to continuously predict and optimize the lidar's wind speed, ensuring the lidar is not obstructed by blades and improving wind speed prediction accuracy. Simultaneously, a backpropagation neural network is used to correct the expected pitch angle increment, enabling real-time adjustment of the pitch angle. This achieves automatic pitch angle control, preventing turbine overspeed and reducing power generation losses.

[0055] Figure 1 A flowchart illustrating a control method for a wind turbine lidar provided in this application embodiment is shown below. Figure 1 As shown, the method includes:

[0056] S10: Obtain the initial estimated wind speed;

[0057] S11: The initial estimated wind speed is filtered by Kalman filtering to correct the initial estimated wind speed and obtain the optimal estimate.

[0058] In specific embodiments, the actual measured wind speed near the hub is often affected by factors such as blade shielding, blade rotation, and clouds, etc., and it is difficult to verify whether the measured wind speed is accurate, and the estimated wind speed is often inaccurate, therefore, the embodiment of the application filters the initial estimated wind speed by Kalman filtering to realize the correction of the initial estimated wind speed.

[0059] Figure 2 Another schematic diagram of the control method of the wind turbine laser radar provided by the embodiment of the application is shown in FIG. 2. Figure 2 After obtaining the initial estimated wind speed Lidar, the initial estimated wind speed Lidar is predicted and corrected by Kalman filtering, and the optimal estimated value is obtained after correction. It should be noted that the Kalman filtering can be said to correct the initial estimated wind speed Lidar, or optimize, or filter, or of course modify, all of which represent the same meaning, that is, to continuously update the initial estimated wind speed Lidar, and improve the accuracy of the estimated wind speed.

[0060] In implementation, when the initial estimated wind speed Lidar is corrected by Kalman filtering, a laser radar wind measurement model is first constructed according to a pre-established reference coordinate system. It should be noted that the laser radar can be a pulsed laser radar or a continuous wave laser radar, which is not limited by the application. Since the continuous wave laser radar can effectively collect wind data under various weather conditions, the continuous wave laser radar is preferred.

[0061] Figure 3 A schematic diagram of a reference coordinate system provided by the embodiment of the application is shown in FIG. 3. Figure 3 As shown in FIG. 3, the embodiment of the application adopts four continuous wave laser radars, the sampling frequency is 4Hz, the laser beam is 30 degrees with the horizontal plane, and 25 degrees with the vertical plane. The pre-established reference coordinate system takes the tower base center as the reference, the downwind direction is the positive direction x-axis of the tower base, and the direction of gravity is the opposite direction z-axis.

[0062] After constructing the laser radar wind measurement model, the laser radar wind measurement model is converted into a discretized state space equation, and the next state wind speed is predicted to obtain the optimal estimated value according to the discretized state space equation and the initial estimated wind speed. Further, in order to avoid potential instability, the optimal estimated value is optimized based on a positive definite matrix, thereby further optimizing the estimated wind speed.

[0063] S12: calculating each optimal estimated value to obtain an expected pitch angle increment;

[0064] S13: correcting the expected pitch angle increment by a BP neural network to obtain a target pitch angle increment;

[0065] S14: superimpose the target pitch angle increment to the given pitch angle increment to output the actual rotating speed.

[0066] In the implementation, the expected pitch angle increment can be obtained by calculating each optimal estimated value through the fitting algorithm with the optimal estimated value as the input. Then, the expected pitch angle increment is corrected through the BP neural network to solve the problem of overspeed of the wind wheel rotating speed.

[0067] In the correction, the neural network training sample is constructed through the preset reconstruction method. In order to improve the correction accuracy and efficiency, after the neural network training sample is constructed, the input layer, the hidden layer and the output layer of the BP neural network are initialized respectively, and then the hidden layer and the output layer of the BP neural network are calculated.

[0068] Further, the weights of the BP neural network are adjusted in the error gradient descent manner, and the threshold values of the hidden layer and the output layer are updated. Finally, the target pitch angle increment is calculated according to the Jacobian matrix after the Jacobian matrix is constructed.

[0069] As shown in Figure 2 , the target pitch angle increment is obtained after the pitch angle increment is corrected through the BP neural network, that is, the feedforward pitch angle increment is obtained. The target pitch angle increment is added to the given pitch angle increment by the variable pitch actuator to output the actual rotating speed. Thus, the control of the wind wheel rotating speed is realized through the continuous correction of the feedforward pitch angle increment.

[0070] The control method of the wind turbine laser radar provided by the embodiment of the application comprises: obtaining an initial estimated wind speed, and filtering the obtained initial estimated wind speed through Kalman filtering to correct the initial estimated wind speed to obtain an optimal estimated value. Then, each optimal estimated value is calculated to obtain an expected pitch angle increment, and the expected pitch angle increment is corrected through a BP neural network to obtain a target pitch angle increment. The target pitch angle increment is superimposed to the given pitch angle increment to output the actual rotating speed. As can be seen, the technical solution provided by the application continuously predicts and optimizes the wind speed of the laser radar through Kalman filtering, so that the laser radar can accurately predict the wind speed even if there is blade shielding and the like, and the wind turbine control effect is improved. In addition, the expected pitch angle increment is corrected through the BP neural network to realize real-time adjustment of the pitch angle increment, that is, to realize control of the variable pitch, avoid overspeed of the wind wheel rotating speed, and reduce the loss of power generation.

[0071] As a preferred embodiment, Figure 4 A flow chart of a control method of a wind turbine laser radar provided by another embodiment of the application is shown in Figure 4 , filtering the initial estimated wind speed through Kalman filtering to correct the initial estimated wind speed to obtain an optimal estimated value comprises:

[0072] S20: Constructing a laser radar wind measurement model according to a pre-established reference coordinate system;

[0073] In implementation, the laser radar mainly includes two types of pulsed laser radar and continuous wave laser radar, both of which use the Doppler shift principle to measure wind speed. The pulsed laser radar is suitable for high-altitude measurement due to high laser energy, while the continuous wave laser radar is used for low-altitude measurement and can effectively collect wind data under various weather conditions. The embodiment of the application adopts four continuous laser radars, the sampling frequency of which is 4Hz, the laser beam forms a 30-degree angle with the horizontal plane and a 25-degree angle with the vertical plane.

[0074] As shown in Figure 3 , the pre-established reference coordinate system takes the tower base center as the reference, the downwind direction is the positive direction x-axis of the tower base, and the direction of gravity is the opposite direction z-axis. The mathematical expression of the constructed laser radar wind measurement model is:

[0075]

[0076]

[0077] Among them, F is the focal length of the laser radar, D is the measurement distance of the laser radar beam, v LOS (D) represents the apparent wind speed at a distance D along the laser beam, K is a normalization constant, used to make the entire weighted function integral to 1, N is the number of laser radar beams, usually 4.

[0078] The laser radar collects Figure 3 the wind speed information of four points in the same plane, and the equivalent wind speed at the distance gate (the plane formed by the four laser beams emitted by the laser radar is called "gate") is calculated as follows:

[0079]

[0080] According to the Taylor frozen turbulence model, the wind speed lead time can be approximately T≈d / v, where T is the time from the gate to the wind wheel surface, d is the distance from the gate to the wind wheel surface, and v is the average wind speed.

[0081] S21: Convert the laser radar wind measurement model into a discretized state space equation;

[0082] After constructing the laser radar wind measurement model, the laser radar wind measurement model is converted into a discretized state space equation, and the specific expression is:

[0083] x k =f(x k )+v k-1

[0084] y k =g(x k)+w k

[0085] where x k is the state equation, which represents the wind speed at the hub estimated by the wind lidar model through measuring the wind speed. f(x k ) is the state transition matrix, v k-1 is the process noise.

[0086] y k is the measurement equation, which is used to describe the system state, including the wind turbine speed, the generator speed, and the wind speed measured by the distance gate of the lidar. g(x k ) is the measurement matrix, and w k is the measurement noise.

[0087] S22: The wind speed of the next state is predicted according to the discretized state space equation and the initial estimated wind speed to obtain an optimal estimation value.

[0088] In the implementation, the process model of the system is used to predict the next state of the system. Assuming that the current state of the system is k, according to the system model, the current state of the system can be predicted based on the last state of the system, and the current state of the system is:

[0089] x k - ≈f(x k-1 )

[0090] P k - =J n (x k-1 )P k-1 J f (x k-1 )+Q k-1 where x k - is the prior state estimation, which is the result predicted by the last state, f(x k-1 ) is the optimal result of the last state, P k - is the prior estimation error covariance, J n (x k-1 ) is the Jacobian matrix of f(x k-1 ), Q is the process noise covariance matrix, and the process noise covariance matrix is:

[0091] Q=E[v k-1 v k-1 T ]

[0092] The wind speed correction formula of the lidar is as follows:

[0093] xk-1 ≈x k - +K k (y k -g(x k ))

[0094] K k =P k - J q T (x k (J) q (x k )P k - J q T (x k )) -1 +R k

[0095] P k =P k - -K k J q (x k )P k -

[0096] J g (x k Let K be the Jacobian matrix of g(x), and K be the Jacobian matrix of g(x). k For Kalman gain, R k To measure the noise covariance matrix, where R k Satisfy the following expression:

[0097] R k T =E[w k w k T ]

[0098] Based on the prediction results of the current state, and by collecting the measured values ​​of the current state, and combining the predicted and measured values, the optimal estimate P of the current state t can be obtained. k .

[0099] In step S22, if Q is a constant, and R... k For real-time updates, the optimal estimate P is... k The difference between two positive definite matrices in the calculation may produce a non-positive definite matrix, leading to numerical instability. To avoid potential instability, the optimal estimate P is... k Further revisions are needed:

[0100] P k =Pk - (I-K k J q (x k ))P k - (I-K k J q (x k )) T +K k R k K k T

[0101] Therefore, the two terms in the equation are always positive definite, so when the sum of two positive definite matrices is a positive definite matrix, P k is also guaranteed to be positive definite.

[0102] The wind turbine laser radar control method provided by the embodiment of the application continuously iteratively updates the covariance and the optimal estimation value through Kalman filtering, so as to predict and optimize the wind speed of the laser radar, and ensure that the laser radar can accurately predict the incoming wind even under the influence of factors such as blade shielding, occasional beam abnormalities and clouds and fog.

[0103] Figure 5 The flow chart of the wind turbine laser radar control method provided by another embodiment of the application is shown in Figure 5 As a preferred embodiment, the target pitch angle increment is obtained by correcting the expected pitch angle increment through the BP neural network, and the method comprises the following steps:

[0104] S30: Constructing a neural network training sample through a preset reconstruction method;

[0105] Figure 6 The schematic diagram of the BP neural network provided by the embodiment of the application is shown in Figure 6 According to the BP neural network, the target function is defined as:

[0106]

[0107] Wherein, N is the total number of samples, m is the number of layers of the output layer, o is the output value, and y is the expected value.

[0108] The neural network training sample can be constructed by using the derivative reconstruction method or the phase space reconstruction method. In order to mine the dynamic characteristics of the entire time series and avoid collecting prior information, the phase space reconstruction method is preferably used for construction.

[0109] The feedforward output time sequence of the laser radar is {x(i)}, i∈[1, n0], the embedding dimension is m0, the delay time is τ, n0 is the number of collection points of the time sequence, and P is the prediction step, so the training sample of the constructed neural network is:

[0110] [x i ,y i ]=[(x(i),x(i+τ),...,x(i+(m0-1)τ)) T ,x(i+(m0-1)τ+P)]

[0111] In order to improve the accuracy and efficiency of the pitch angle increment correction, after the neural network training sample is constructed, the input layer, the hidden layer and the output layer of the BP neural network are initialized respectively. Specifically, the BP neural network is divided into an input layer, a hidden layer and an output layer, the number of layers of each layer is initialized as n, l, m layers in turn, and the number of neurons of each layer is i, j, k. The weight value from the input layer to the hidden layer is ω ij , the weight value from the hidden layer to the output layer is ω jk , the input layer vector is a=[a1,..., an], the hidden layer vector is b=[b1,..., bl], and the output layer vector is c=[c1,..., cm].

[0112] S31: calculating the hidden layer and the output layer in the BP neural network;

[0113] According to the initialized BP neural network, the hidden layer and the output layer are calculated respectively:

[0114]

[0115]

[0116] Wherein, x is an input node variable, and f is an excitation function of the hidden layer. The excitation function is a Sigmoid function:

[0117] f(s)=(a+exp(-cs)) -1

[0118] S is the input value of the neuron, f(s) is the output value of the neuron, the excitation function is a single-input single-output monotonically increasing function, and the output response is only related to the current input value.

[0119] S32: adjusting the weight value of the BP neural network in the manner of error gradient descent, and updating the threshold value of the hidden layer and the output layer;

[0120] The weight value of the BP neural network is updated, that is, the training adjustment is performed in the manner of error gradient descent:

[0121]

[0122]

[0123] wherein η is a learning rate and β is a momentum factor.

[0124] Further, the threshold values of the hidden layer and the output layer are updated, including:

[0125]

[0126] c k (t+1)=c k (t)+(y k -o k )

[0127] S33: After constructing the Jacobian matrix, the target pitch angle increment is calculated according to the Jacobian matrix.

[0128] It should be noted that, as shown in Figure 2 , the target pitch angle increment is the feedforward pitch angle increment, and when calculating the feedforward pitch angle increment, the first input of the BP neural network is taken as βFF(t), which is the pitch angle increment calculated after the Kalman filtering of the laser radar, as the laser radar pitch angle before correction. And Δβ FF (t)=β FF (t)-β FF (t-1) is taken as the subsequent input, , wherein w i represents the weighting coefficient of the i th delay signal, i=[0,N-1] constitute the Jacobian matrix, and the Jacobian matrix can provide the FIR parameter adjustment algorithm, and the calculation formula is as follows:

[0129]

[0130]

[0131] Therefore, the target pitch angle increment can be calculated:

[0132]

[0133] Further, the target pitch angle increment is superimposed on the given pitch angle increment, that is, the feedforward pitch angle increment is superimposed on the given pitch angle increment, and the actual rotating speed can be obtained.

[0134] The control method of the wind turbine laser radar provided by the embodiment of the application controls the variable pitch above the rated wind speed through the BP neural network, avoids the overspeed of the wind wheel, and reduces the loss of power generation.

[0135] In the above embodiments, the control method of the wind turbine laser radar is described in detail, and the application also provides a corresponding embodiment of a control device of the wind turbine laser radar. It should be noted that the embodiments of the device part are described from two angles, one is based on the functional module angle, and the other is based on the hardware structure angle.

[0136] Figure 7 A structural diagram of a control device of a wind turbine laser radar provided by an embodiment of the application is shown in FIG. 1, which includes: Figure 7

[0137] The acquisition module 10 is configured to acquire an initial estimated wind speed.

[0138] The filtering module 11 is configured to filter the initial estimated wind speed by Kalman filtering to correct the initial estimated wind speed to obtain an optimal estimation value.

[0139] The calculation module 12 is configured to calculate each optimal estimation value to obtain an expected pitch angle increment.

[0140] The correction module 13 is configured to correct the expected pitch angle increment by a BP neural network to obtain a target pitch angle increment.

[0141] The superposition module 14 is configured to superimpose the target pitch angle increment on a given pitch angle increment to output an actual rotating speed.

[0142] Since the embodiments of the device part correspond to the embodiments of the method part, the embodiments of the device part are described in the description of the embodiments of the method part, which will not be described here.

[0143] The control device of the wind turbine laser radar provided by the embodiment of the application includes: acquiring an initial estimated wind speed, and filtering the acquired initial estimated wind speed by Kalman filtering to correct the initial estimated wind speed to obtain an optimal estimation value. Then, each optimal estimation value is calculated to obtain an expected pitch angle increment, and the expected pitch angle increment is corrected by a BP neural network to obtain a target pitch angle increment. The target pitch angle increment is superimposed on a given pitch angle increment to output an actual rotating speed. As can be seen, the technical solution provided by the application continuously predicts and optimizes the wind speed of the laser radar by Kalman filtering, so that the laser radar can accurately predict the wind speed even if there is blade shielding and other influences, and the wind turbine control effect is improved. In addition, the expected pitch angle increment is corrected by the BP neural network to realize real-time adjustment of the pitch angle increment, that is, to realize control of the pitch, to avoid overspeed of the wind wheel, and to reduce the loss of power generation.

[0144] Figure 8 A structural diagram of a control device of a wind turbine laser radar provided by another embodiment of the application is shown in FIG. 2, which includes: Figure 8 ​As shown, the control device of the wind turbine laser radar comprises a memory 20 for storing a computer program;

[0145] A processor 21 is configured to implement the steps of the control method of the wind turbine laser radar as mentioned in the above embodiments when executing the computer program.

[0146] The control device of the wind turbine laser radar provided by the embodiment can include, but is not limited to, a smart phone, a tablet computer, a notebook computer, a desktop computer, etc.

[0147] The processor 21 can include one or more processing cores, such as a 4-core processor, an 8-core processor, etc. The processor 21 can be implemented in at least one of a hardware form of a Digital Signal Processor (DSP), a Field-Programmable Gate Array (FPGA), and a Programmable Logic Array (PLA). The processor 21 can also include a main processor and a coprocessor. The main processor is a processor for processing data in a wake-up state, also known as a Central Processing Unit (CPU). The coprocessor is a low-power processor for processing data in a standby state. In some embodiments, the processor 21 can be integrated with a Graphics Processing Unit (GPU) for rendering and drawing the content to be displayed on the display screen. In some embodiments, the processor 21 can further include an Artificial Intelligence (AI) processor for processing machine learning-related computing operations.

[0148] The memory 20 can include one or more computer-readable storage media, which can be non-transitory. The memory 20 can further include a high-speed random access memory, and a non-volatile memory such as one or more disk storage devices, flash storage devices. In the embodiment, the memory 20 is at least used to store the following computer program 201, wherein the computer program is loaded and executed by the processor 21, and can implement the related steps of the control method of the wind turbine laser radar disclosed in any of the preceding embodiments. In addition, the resources stored in the memory 20 can further include an operating system 202 and data 203, etc., and the storage mode can be temporary storage or permanent storage. The operating system 202 can include Windows, Unix, Linux, etc. The data 203 can include, but is not limited to, related data involved in the control method of the wind turbine laser radar, etc.

[0149] In some embodiments, the control device of the wind turbine laser radar can further include a display screen 22, an input / output interface 23, a communication interface 24, a power supply 25, and a communication bus 26.

[0150] Those skilled in the art can understand that the structure shown in the above Figure 8 The structure shown in the above does not constitute a limitation on the control device of the wind turbine laser radar, and can include more or fewer components than those shown in the figure.

[0151] The control device of the wind turbine laser radar provided by the embodiments of the present application includes a memory and a processor. When the processor executes a program stored in the memory, the following method can be implemented: a control method of the wind turbine laser radar.

[0152] The control device of the wind turbine laser radar provided by the embodiments of the present application continuously predicts and optimizes the wind speed of the laser radar through Kalman filtering, so as to ensure that the laser radar can accurately predict the wind speed even if there is a blade shielding and the like, and to improve the wind turbine control effect. In addition, the expected pitch angle increment is corrected through the BP neural network, so as to realize real-time adjustment of the pitch angle increment, that is, to realize control of the variable pitch, to avoid overspeed of the wind wheel, and to reduce the loss of power generation.

[0153] Finally, the present application also provides an embodiment corresponding to a computer readable storage medium. The computer readable storage medium stores a computer program. When the computer program is executed by a processor, the steps recorded in the above method embodiments are implemented.

[0154] It can be understood that if the method in the above embodiments is implemented in the form of a software function unit and sold or used as an independent product, it can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or say the part that contributes to the prior art or the whole or part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium, and executes all or part of the steps of the methods described in the embodiments of the present application. The foregoing storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0155] The above introduces in detail the method, device and medium for controlling the wind turbine laser radar. The embodiments in the specification are described in a progressive manner, and each embodiment mainly explains the difference from other embodiments. The same or similar parts of each embodiment can be understood by referring to each other. For the device disclosed in the embodiments, since it corresponds to the method disclosed in the embodiments, the description is relatively simple, and the related parts can be understood by referring to the method part. It should be pointed out that, for those skilled in the art, without departing from the principles of the present application, the present application can be improved and modified in several ways, and these improvements and modifications also fall within the protection scope of the claims of the present application.

[0156] It should also be noted that in this specification, the relationship terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between the entities or operations. Moreover, the terms "include", "contain" or any other variants thereof are intended to cover non-exclusive inclusion, so that the process, method, article or device including a series of elements not only includes those elements, but also includes other elements not explicitly listed or inherent to such process, method, article or device. Without further limitation, the element defined by the statement "including a" does not exclude the presence of other identical elements in the process, method, article or device including the element.

Claims

1. A control method for a wind turbine lidar, characterized in that, include: Obtain the initial estimated wind speed; The initial estimated wind speed is filtered by Kalman filtering to correct the initial estimated wind speed and obtain the optimal estimate; The expected pitch angle increment is obtained by calculating each of the optimal estimates. The target pitch angle increment is obtained by correcting the expected pitch angle increment using a BP neural network. The target pitch angle increment is superimposed on the given pitch angle increment to output the actual rotational speed; wherein: The step of filtering the initial estimated wind speed using Kalman filtering to correct the initial estimated wind speed and obtain the optimal estimate includes: A lidar wind measurement model is constructed based on a pre-established reference coordinate system; The lidar wind measurement model is transformed into a discretized state-space equation; The optimal estimate is obtained by predicting the wind speed of the next state based on the discretized state-space equation and the initial estimated wind speed; the optimal estimate is then optimized based on a positive definite matrix. The step of constructing a lidar wind measurement model based on a pre-established reference coordinate system includes: A reference coordinate system is pre-established, with the center of the tower base as the reference, the downwind direction as the positive x-axis of the tower base, and the direction of gravity as the opposite z-axis. A lidar wind measurement model is constructed based on the aforementioned reference coordinate system. The expression of the lidar wind measurement model is as follows: ； Where F represents the focal length of the lidar, and D represents the measurement distance of the lidar beam. The line-of-sight wind speed is represented at a distance D along the laser beam, K represents a normalization constant used to make the integral of the entire weighted function equal to 1, and N represents the number of lidar beams, which is usually 4. This indicates the output velocity of the lidar wind measurement model; The step of transforming the lidar wind measurement model into a discretized state-space equation includes: Based on expression and The lidar wind measurement model is transformed into a discretized state-space equation; where, The state equation is... This indicates that the lidar wind measurement model estimates the wind speed at the wheel hub by measuring the wind speed. Here is the state transition matrix. This is process noise; For the measurement equation, Used to describe the system state. The wind turbine speed, generator speed, and range gate wind speed measured by lidar are the state variables of the Kalman filter observer; For the measurement matrix, For measuring noise; k represents the current system state, and k-1 represents the previous system state; The step of correcting the expected pitch angle increment using a BP neural network to obtain the target pitch angle increment includes: Neural network training samples are constructed using a pre-defined reconstruction method; The hidden layers and output layer of the BP neural network are calculated; The weights of the BP neural network are adjusted using the error gradient descent method, and the thresholds of the hidden layer and the output layer are updated. After constructing the Jacobian matrix, the target pitch angle increment is calculated based on the Jacobian matrix; The lidar is a continuous wave lidar.

2. The control method for wind turbine lidar according to claim 1, characterized in that, The process further includes, prior to the computation of the hidden and output layers in the BP neural network: The input layer, hidden layer, and output layer of the BP neural network are initialized respectively.

3. The control method for wind turbine lidar according to claim 2, characterized in that, The preset reconstruction method is the phase space reconstruction method.

4. A control device for a wind turbine lidar, characterized in that, include: The acquisition module is used to obtain the initial estimated wind speed; The filtering module is used to filter the initial estimated wind speed using Kalman filtering to correct the initial estimated wind speed and obtain the optimal estimate. The calculation module is used to calculate the expected pitch angle increment from each of the optimal estimates; The correction module is used to correct the expected pitch angle increment using a BP neural network to obtain the target pitch angle increment; The superposition module is used to superimpose the target pitch angle increment onto a given pitch angle increment to output the actual rotational speed; The filtering module is used to construct a lidar wind measurement model based on a pre-established reference coordinate system; convert the lidar wind measurement model into a discretized state-space equation; predict the wind speed of the next state based on the discretized state-space equation and the initial estimated wind speed to obtain the optimal estimate; and optimize the optimal estimate based on a positive definite matrix. The step of constructing a lidar wind measurement model based on a pre-established reference coordinate system includes: A reference coordinate system is pre-established, with the center of the tower base as the reference, the downwind direction as the positive x-axis of the tower base, and the direction of gravity as the opposite z-axis. A lidar wind measurement model is constructed based on the aforementioned reference coordinate system. The expression of the lidar wind measurement model is as follows: ； Where F represents the focal length of the lidar, and D represents the measurement distance of the lidar beam. The line-of-sight wind speed is represented at a distance D along the laser beam, K represents a normalization constant used to make the integral of the entire weighted function equal to 1, and N represents the number of lidar beams, which is usually 4. This indicates the output velocity of the lidar wind measurement model; The step of transforming the lidar wind measurement model into a discretized state-space equation includes: Based on expression and The lidar wind measurement model is transformed into a discretized state-space equation; where, The state equation is... This indicates that the lidar wind measurement model estimates the wind speed at the wheel hub by measuring the wind speed. Here is the state transition matrix. This is process noise; For the measurement equation, Used to describe the system state. The wind turbine speed, generator speed, and range gate wind speed measured by lidar are the state variables of the Kalman filter observer; For the measurement matrix, For measuring noise; k represents the current system state, and k-1 represents the previous system state; The step of correcting the expected pitch angle increment using a BP neural network to obtain the target pitch angle increment includes: Neural network training samples are constructed using a pre-defined reconstruction method; The hidden layers and output layer of the BP neural network are calculated; The weights of the BP neural network are adjusted using the error gradient descent method, and the thresholds of the hidden layer and the output layer are updated. After constructing the Jacobian matrix, the target pitch angle increment is calculated based on the Jacobian matrix; The lidar is a continuous wave lidar.

5. A control device for a wind turbine lidar, characterized in that, Includes memory used to store computer programs; A processor is configured to execute the computer program to implement the steps of the control method for a wind turbine lidar as described in any one of claims 1 to 3.

6. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores a computer program that, when executed by a processor, implements the steps of the control method for the wind turbine lidar as described in any one of claims 1 to 3.

Images