A method for reducing wind farm unit misoperation based on short-term wind speed prediction

Abstract

The application discloses a method for reducing wind turbine misoperation of a wind farm based on short-term wind speed prediction, comprising the following steps: step one, obtaining layout information of the whole wind farm, obtaining wind tower positions of the wind farm and extracting historical wind speed data recorded by the wind towers; step two, based on historical measured wind speed information of the wind tower positions of the wind farm and natural incoming flow wind speed at the current moment, predicting the natural incoming flow wind speed at the future moment by using a prediction model pre-built by an artificial intelligence modeling method, and obtaining short-term predicted wind speed u t,0 ; step three, according to the predicted natural incoming flow wind speed u t,0 at the wind tower positions, combining the layout information of the whole wind farm, and based on a wake model with delay, calculating the effective incoming flow wind speed u(T,j) at each wind turbine considering the delay; and step four, based on the predicted effective incoming flow wind speed at the hub positions of each wind turbine, achieving the optimization purpose by means of fan start-stop control and variable pitch control.

Description

Technical Field

[0001] This invention belongs to the field of wind farm automatic control and relates to a method for reducing wind farm unit malfunctions based on short-term wind speed prediction. Background Technology

[0002] To address the energy gap and related problems caused by the traditional fossil fuel crisis, scientists have begun to focus on the development and utilization of clean energy. Wind energy, with its vast reserves and clean, pollution-free characteristics, has become a key area for development. However, wind energy is an intermittent resource, highly dependent on environmental and meteorological conditions. Its high forecasting difficulty and poor stability and continuity make it unsuitable for grid connection. Improving the accuracy of wind power forecasting is crucial for extending wind turbine lifespan and ensuring the safe and stable operation of the power system. This makes simultaneous prediction of real-time wind speeds for each wind turbine in a wind farm, along with start-up, shutdown, and pitch control, a key research area in wind farms. However, the large delays and strong uncertainties in wind farm operations present numerous challenges to reducing the normal operation of wind turbine units.

[0003] Influenced by natural geographical environment and other factors, the complex conditions within the atmospheric boundary also lead to the fluctuation and randomness of wind speed; at the same time, in order to improve the efficiency of wind energy utilization, wind turbine units must maintain a certain distance to reduce the wake impact of upstream wind turbines on downstream wind turbines, which results in a certain delay time for the wake of upstream units to flow to downstream units.

[0004] The uncertainty of wind speed and the interaction between wakes in wind farms lead to the ineffective and unstable operation of wind turbines, mainly due to two situations: low wind speed and high wind speed. 1) Low wind speeds cause frequent start-stops of wind turbines, shortening the lifespan of the inverters inside the turbine; affecting the communication slip rings of the turbine; shortening the lifespan of the pitch motor; affecting the load and shortening the service life of the turbine. 2) While fixed-pitch control is simple and reliable, fixed-pitch blades are heavier (compared to variable-pitch blades), and the control effect depends on the aerodynamic characteristics of the blades, resulting in a significant drop in output power at high wind speeds. Variable-pitch control has many advantages: by changing the installation angle, larger starting and braking torques can be obtained; during grid-connected operation, the installation angle can be adjusted according to wind speed changes to control the aerodynamic torque of the rotor, thereby stabilizing the turbine's output power; when the wind speed is particularly high, the blades can be adjusted to a feathered position, greatly improving the stress on the blades and tower; the blades of variable-pitch wind turbines are relatively light, and the generator's design overload capacity is smaller.

[0005] Therefore, how to achieve stable control of wind farms under delayed conditions and strong uncertainties has become the key to reducing the misoperation of wind farm units. Summary of the Invention

[0006] Purpose of the invention: In view of the problems and deficiencies of the prior art, the purpose of this invention is to provide a method for reducing the malfunction of wind farm units based on short-term wind speed prediction, which can simultaneously handle the uncertainty of wind speed and the mutual influence of wake, and ensure the safe and stable operation of wind turbines.

[0007] Technical solution: A method for reducing wind farm turbine malfunctions based on short-term wind speed forecasting, comprising the following steps:

[0008] Step 1: Obtain the layout information of the entire wind farm, find the location of the wind farm's meteorological towers, and extract the historical wind speed data recorded by the meteorological towers;

[0009] Step 2: Based on historical measured wind speed information at the wind farm's anemometer tower locations and the current natural incoming wind speed, a pre-built prediction model using artificial intelligence modeling methods is used to predict the future natural incoming wind speed, obtaining the short-term predicted wind speed u. t,0 ;

[0010] Step 3: Based on the predicted natural incoming wind speed u at the location of the meteorological tower t,0 Based on the layout information of the entire wind farm and the wake model with delay, the effective incoming wind speed u(T,j) at the rotor of each unit considering the delay is calculated.

[0011] Step 4: Based on the predicted effective incoming wind speed at each wind turbine hub, the optimization is achieved by controlling the start and stop of the wind turbines if the wind speed does not exceed the set minimum threshold, and by controlling the pitch if the wind speed exceeds the set maximum threshold.

[0012] Furthermore, in step one, obtaining the layout information of the entire wind farm specifically involves: modeling the position of each wind turbine in the wind farm, and finding the relative position of the meteorological tower relative to the entire wind farm; the historical wind speed data recorded by the meteorological tower, including wind speed and wind direction information, and removing unreasonable items from the wind speed information.

[0013] Furthermore, the specific steps of step two are as follows:

[0014] 1) Set 95% of the historical wind speed data from the wind farm's wind measurement towers as the training dataset and the remaining 5% as the test dataset. Train an LSTM neural network to obtain a prediction model.

[0015] 2) Use the last data point in the training dataset as input for wind speed prediction, and use the established prediction model to predict wind speed.

[0016] Furthermore, the specific calculation steps for step three are as follows:

[0017] Step 1) Obtain the natural incoming wind speed u at the location of the wind measurement tower over a future period of time through Step 2. t,0 The wind turbine with the smallest projected distance from the wind measuring tower under the wind direction is considered to have the same wind speed as the wind measuring tower, denoted as u(t,i). The wind speed u(t,j) at the downstream wind turbine location is calculated based on the engineering wake model.

[0018]

[0019] In the formula, a is the axial induction factor; u(t,i) is the wind speed at the upstream wind turbine hub position at time t; u(t,j) is the wind speed at the downstream wind turbine hub position; x is the projection of the distance between the upstream and downstream wind turbines in the wind direction; D is the rotor diameter; r is the rotor radius; r0 is the initial wake radius behind the wind turbine; and k is the wake attenuation coefficient.

[0020] Step 2) Calculate the wake delay time τ from the i-th unit to the j-th unit. ij :

[0021]

[0022] In the formula, the propagation time between the i-th wind turbine and the j-th wind turbine is calculated by integration; u(t,i) is the wind speed at the hub position of the upstream wind turbine at time t; τ ij It is the propagation time between the two wind turbines;

[0023] Step 3) Taking propagation time into account, obtain the wind speed u(T,j) at time T at the height of the downstream wind turbine hub:

[0024]

[0025] In the formula, a is the axial induction factor; x is the projection of the distance between the upstream and downstream wind turbines under the wind direction; r0 is the initial wake radius behind the wind turbine; k is the wake attenuation coefficient; T is the time when the downwind propagates from the upstream wind turbine to the downstream wind turbine at time t; u(t,i) is the wind speed at the hub position of the upstream wind turbine at time t; u(T,i) is the wind speed at the height of the hub of the downstream wind turbine at time T.

[0026] Step 4) Considering the influence of the wake, calculate the weighting factor β for the wake correction;

[0027]

[0028] In the formula, β represents the weighting factor for the wake correction; A ij r represents the projected area of ​​the wake region of the i-th upstream wind turbine onto the rotation plane of the j-th downstream wind turbine; j Let be the radius of the rotor of the j-th wind turbine;

[0029]

[0030]

[0031]

[0032] In the formula, r x d is the wake radius when the i-th wind turbine reaches the location of the j-th wind turbine; ij r is the distance between the centers of the i-th and j-th wind turbine hubs under vertical wind direction. j It is the radius of the j-th downstream wind turbine rotor; A ij θ1 and θ2 represent the projected area of ​​the wake region of the upstream i-th wind turbine on the rotation plane of the downstream j-th wind turbine rotor; θ1 and θ2 are the center angles between the i-th and j-th wind turbine rotors and the wake influence section, respectively.

[0033] Step 5) By superimposing the wind speeds arriving at the wind turbine at the same time using a sum-of-squares model, the wind speed at the j-th downstream wind turbine hub position at that time is obtained:

[0034]

[0035] In the formula, N is the number of upstream wind turbines; n is the nth upstream wind turbine; u(T,j n Let u(t,i) be the wake of the upstream n-th wind turbine at time T, located at the downstream j-th wind turbine. n u(T,j) is the wind speed at the hub position of the nth upstream wind turbine at time t; u(T,j) is the wind speed at the height of the hub of the jth downstream wind turbine after stacking at time T; u(t,0) is the wind speed at the meteorological tower at time t; β is the weighting factor for the wake correction.

[0036] Furthermore, in step four:

[0037] 1) Optimization is achieved through fan start / stop control when the wind speed does not exceed the set minimum threshold. Specifically:

[0038] If the incoming wind speed remains fluctuating around the cut-in wind speed for a period of time, and does not exceed 120% of the cut-in wind speed, the wind turbine should be shut down through the control system to avoid frequent start-stop of the wind turbine.

[0039] 2) When the wind speed is higher than the set maximum threshold, the optimization purpose is achieved by variable pitch control; variable pitch adjustment refers to rotating the blades along the longitudinal axis of the blades, and through fine angle changes, making the blades rotate in the feathering direction, changing the blade speed, and realizing the power control of the unit.

[0040] This invention provides a processor for operating according to instructions to execute a method for reducing wind farm unit malfunctions based on short-term wind speed forecasts.

[0041] The present invention provides a storage medium storing a computer program thereon, the storage medium being used to store instructions for a method of reducing wind farm unit malfunctions based on short-term wind speed forecasting; when the computer program is executed by a processor, it implements the method of reducing wind farm unit malfunctions based on short-term wind speed forecasting.

[0042] Compared with the prior art, the present invention has the following advantages:

[0043] 1) The control-oriented quasi-steady-state model of wind farm established in this invention is based on the steady-state background flow field with added delay processing, which can further improve the accuracy of the wake model and realize the rapid estimation of the dynamic flow field of wind farm.

[0044] 2) This invention uses the natural incoming wind speed to calculate the delay time, and combines it with the steady-state flow field model to obtain the quasi-steady-state flow field model. It retains the convenient wake coupling relationship of the traditional steady-state model, and also considers the dynamic characteristics of the wake delay. Thus, it can approximately simulate the main dynamic characteristics of the wind farm, and can be applied to the rapid online calculation of the wake of large-scale wind farms or even wind power bases.

[0045] 3) Due to the strong nonlinearity of wind speed and the need for rapid online solutions, an artificial intelligence modeling method is used for short-term wind speed prediction in wind farms. This method is a data-driven modeling approach that can fully extract nonlinear information from the data and solve the problem efficiently and quickly online.

[0046] 4) The method of the present invention can reduce the start-up and shutdown frequency of wind turbine units, extend the life of the frequency converter inside the wind turbine, extend the life of the pitch motor, and extend the service life of the wind turbine. Attached Figure Description

[0047] Figure 1 This is a flowchart of the method of the present invention. Detailed Implementation

[0048] The technical solution of the present invention will be described in detail below with reference to the accompanying drawings, but the scope of protection of the present invention is not limited to the embodiments described.

[0049] This invention focuses on two aspects: establishing a more accurate wake model and adopting more advanced control methods. Regarding wake models, in engineering, empirical formulas such as Jensen, Gaussian, and Frandson are mainly used, but these are mostly steady-state empirical models. Steady-state models cannot consider the delay characteristics of the flow field, thus reducing the accuracy of wind farm models. Due to the fluctuation and randomness of wind speed, frequent start-stop operations of wind turbines and constant-pitch stall control bring a series of adverse effects. Therefore, how to effectively combine wake models with delay characteristics to control wind turbine units has become a key research focus for reducing wind farm unit malfunctions.

[0050] Example 1

[0051] like Figure 1 As shown, a method for reducing wind turbine malfunctions based on short-term wind speed forecasting includes the following steps:

[0052] Step 1: Obtain the layout information of the entire wind farm, find the location of the wind farm's meteorological towers, and extract the historical wind speed data recorded by the meteorological towers;

[0053] In step 1), the layout information of the entire wind farm is obtained, specifically: the position of each wind turbine in the wind farm is modeled, and the relative position of the meteorological tower relative to the entire wind farm is found; the historical wind speed data recorded by the meteorological tower, including wind speed and wind direction information, is used to remove unreasonable items in the wind speed information.

[0054] Step 2: Based on historical measured wind speed information at the wind farm's anemometer tower locations and the current natural incoming wind speed, a pre-built prediction model using artificial intelligence modeling methods is used to predict the future natural incoming wind speed, obtaining the short-term predicted wind speed u. t,0 ;

[0055] Short-term wind speed forecast refers to the wind speed in the next tens of minutes or even several hours.

[0056] Long Short-Term Memory (LSTM) neural networks can solve the vanishing and exploding gradient problems. Furthermore, due to the strong discontinuity and non-stationarity of wind speed at different times, and its poor linearity, LSTM can be a good approach for wind speed prediction. Compared to conventional recurrent neural networks, LSTM adds three special gate functions: forget gate, memory gate, and output gate. This gives each LSTM unit the special ability to delete or add erroneous information to the module. The gate mechanism supported by LSTM can effectively suppress the data flow in the hidden layer that fluctuates with the clock, while making the structure easier to optimize. LSTM uses a "gate" structure to cancel or strengthen the information in "cell associations," thus retaining important information and deleting unimportant information.

[0057]

[0058] In the formula: tanh and σ are the hyperbolic tangent function and the sigmoid function, respectively; x t and h t These are the input and hidden layer output at time t, respectively. It is the input state of the memory unit; W f W i W c W o These are the weight matrices for the forget gate, input gate, cell state, and output gate, respectively; f t i t C t O t These are the forget gate, input gate, cell state, and output gate at time t; b f b i b c b o These are the bias terms for the forget gate, input gate, cell state, and output gate, respectively.

[0059] 95% of the historical wind speed data from the wind farm's anemometers was used as the training dataset, and the remaining 5% was used as the test dataset. An LSTM neural network was trained to derive a prediction model. The last data point in the training set was then used as input for wind speed prediction, and the constructed prediction model was used to predict wind speed.

[0060] Step 3: Based on the predicted natural incoming wind speed u at the location of the meteorological tower t,0 Based on the layout information of the entire wind farm and the wake model with delay, the effective incoming wind speed u(T,j) at the rotor of each unit considering the delay is calculated.

[0061] The specific calculation steps for step three are as follows:

[0062] Step 1) Obtain the natural incoming wind speed u at the location of the wind measurement tower over a future period of time through Step 2. t,0 The wind turbine with the smallest projected distance from the wind measuring tower under the wind direction is considered to have the same wind speed as the wind measuring tower, denoted as u(t,i). The wind speed u(t,j) at the downstream wind turbine location is calculated based on the engineering wake model.

[0063] In engineering, the Jensen model is generally used for wake models:

[0064]

[0065] In the formula, a is the axial induction factor; u(t,i) is the wind speed at the upstream wind turbine hub position at time t; u(t,j) is the wind speed at the downstream wind turbine hub position; x is the projection of the distance between the upstream and downstream wind turbines in the wind direction; D is the rotor diameter; r is the rotor radius; r0 is the initial wake radius behind the wind turbine; and k is the wake attenuation coefficient.

[0066] Step 2) Calculate the wake delay time τ from the i-th unit to the j-th unit using the engineering wake model. ij :

[0067]

[0068] In the formula, the propagation time between the i-th wind turbine and the j-th wind turbine is calculated by integration; u(t,i) is the wind speed at the hub position of the upstream wind turbine at time t; τ ij It is the propagation time of the two wind turbines.

[0069] Step 3) Taking into account the propagation time, obtain the wind speed u(T,j) at time T at the height of the downstream wind turbine hub;

[0070]

[0071] In the formula, a is the axial induction factor; x is the projection of the distance between the upstream and downstream wind turbines under the wind direction; r0 is the initial wake radius behind the wind turbine; k is the wake attenuation coefficient; T is the time when the downwind propagates from the upstream wind turbine to the downstream wind turbine; u(t,i) is the wind speed at the hub position of the upstream wind turbine at time t; u(T,i) is the wind speed at the hub height of the downstream wind turbine at time T.

[0072] Step 4) Due to the irregular arrangement of wind turbines in the wind farm, a wind turbine may be affected not only by the wake of the wind turbine directly in front of it upstream, but also by the wind turbines diagonally above or below it. Therefore, the influence of the offset wake must be considered.

[0073]

[0074] In the formula, β represents the weighting factor for the wake correction; A ij r represents the projected area of ​​the wake region of the i-th upstream wind turbine onto the rotation plane of the j-th downstream wind turbine; j Let be the radius of the rotor of the j-th wind turbine.

[0075]

[0076]

[0077]

[0078] In the formula, rx d is the wake radius when the i-th wind turbine reaches the location of the j-th wind turbine; ij r is the distance between the centers of the i-th and j-th wind turbine hubs under vertical wind direction. j It is the radius of the j-th downstream wind turbine rotor; A ij θ1 and θ2 represent the projected area of ​​the wake region of the upstream i-th wind turbine on the rotation plane of the downstream j-th wind turbine rotor; θ1 and θ2 are the center angles between the i-th and j-th wind turbine rotors and the wake influence section, respectively.

[0079] Step 5) By superimposing the wind speeds arriving at the wind turbine at the same time using the sum-of-squares model, the wind speed at the j-th downstream wind turbine hub position at that time is obtained.

[0080] Wake superposition model: Sum of squares (SS) model:

[0081]

[0082] In the formula, N is the number of upstream wind turbines; n is the nth upstream wind turbine; u(T,j n Let u(t,i) be the wake of the upstream n-th wind turbine at time T, located at the downstream j-th wind turbine. n u(T,j) is the wind speed at the hub position of the nth upstream wind turbine at time t; u(T,j) is the wind speed at the height of the hub of the jth downstream wind turbine after stacking at time T; u(t,0) is the wind speed at the meteorological tower at time t; β is the weighting factor for the wake correction.

[0083] Step 4: Based on the predicted effective incoming wind speed at each wind turbine hub, the optimization is achieved by controlling the start and stop of the wind turbine when the wind speed is below the set minimum threshold, and by controlling the pitch when the wind speed is above the set maximum threshold.

[0084] 1) Optimization is achieved by controlling the start and stop of the fan when the wind speed does not exceed the set minimum threshold:

[0085] When the wind speed is low, that is, when the wind speed fluctuates around the cut-in wind speed and does not exceed 120% of the cut-in wind speed, the fan start-up and shutdown control shall be implemented.

[0086] The start-up and shutdown status of the wind turbine can be determined by combining the incoming and outgoing wind speeds. If the incoming wind speed fluctuates around the incoming wind speed for the next few tens of minutes or even several hours (e.g., two hours), and does not exceed 120% of the incoming wind speed, this will cause the wind turbine to start and stop frequently. The wind turbine should be shut down by issuing a command through the control system to avoid frequent start-up and shutdown.

[0087] 2) When the wind speed is high (above 8m / s (10m / s for some models)), pitch control is performed.

[0088] The output power of a wind turbine increases with wind speed. However, at excessively high wind speeds, if the turbine blades are not retracted, the load on the turbine will be extremely high, potentially causing serious consequences. The blades are fully deployed at wind speeds of around 3-8 m / s; when the wind speed exceeds 8 m / s (10 m / s for some models), the blades feather, reducing the turbine load. Most wind turbines will feather completely at wind speeds above 20 m / s to minimize the turbine load.

[0089] Variable pitch control refers to rotating the blades along their longitudinal axis and making precise angle changes to make the blades rotate in the feathering direction, thereby changing the blade speed and achieving power control of the unit. This process is often started after the unit reaches its rated power to control the wind turbine's energy absorption and maintain a certain output power.

[0090] This embodiment provides a processor and a storage medium. The storage medium is used to store instructions; the processor is used to perform operations according to the instructions to execute the above-described method. The storage medium stores a computer program thereon, which, when executed by the processor, implements the above-described method.

[0091] As described above, although the invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the invention itself. Various changes in form and detail may be made without departing from the spirit and scope of the invention as defined in the appended claims.

Claims

1. A method for reducing wind farm turbine malfunctions based on short-term wind speed forecasting, characterized in that, Includes the following steps: Step 1: Obtain the layout information of the entire wind farm, find the location of the wind farm's meteorological towers, and extract the historical wind speed data recorded by the meteorological towers; Step 2: Based on historical measured wind speed information at the wind farm's anemometer tower locations and the current natural incoming wind speed, a pre-built prediction model using artificial intelligence modeling methods is used to predict the future natural incoming wind speed, obtaining the short-term predicted wind speed u. t,0 ; Step 3: Based on the predicted natural incoming wind speed u at the location of the meteorological tower t,0 Based on the layout information of the entire wind farm and the wake model with delay, the effective incoming wind speed u(T,j) at the rotor of each unit considering the delay is calculated. Step 4: Based on the predicted effective incoming wind speed at each wind turbine hub, the optimization is achieved by controlling the start and stop of the wind turbines when the wind speed does not exceed the set minimum threshold, and by controlling the pitch when the wind speed exceeds the set maximum threshold. The specific calculation steps for step three are as follows: Step 1) Obtain the natural incoming wind speed u at the location of the wind measurement tower over a future period of time through Step 2. t,0 The wind turbine with the smallest projected distance from the wind measuring tower under the wind direction is considered to have the same wind speed as the wind measuring tower, denoted as u(t,i). The wind speed u(t,j) at the downstream wind turbine location is calculated based on the engineering wake model. ； In the formula, a is the axial induction factor; u(t,i) is the wind speed at the upstream wind turbine hub position at time t; u(t,j) is the wind speed at the downstream wind turbine hub position; x is the projection of the distance between the upstream and downstream wind turbines in the wind direction; D is the rotor diameter; r is the rotor radius; r0 is the initial wake radius behind the wind turbine; and k is the wake attenuation coefficient. Step 2) Calculate the wake delay time from unit i to unit j. : ； In the formula, the propagation time between the i-th wind turbine and the j-th wind turbine is calculated by integration; u(t,i) is the wind speed at the hub position of the upstream wind turbine at time t; It is the propagation time between the two wind turbines; Step 3) Taking propagation time into account, obtain the wind speed u(T,j) at time T at the height of the downstream wind turbine hub: ； In the formula, a is the axial induction factor; x is the projection of the distance between the upstream and downstream wind turbines under the wind direction; r0 is the initial wake radius behind the wind turbine; k is the wake attenuation coefficient; T is the time when the downwind propagates from the upstream wind turbine to the downstream wind turbine at time t; u(t,i) is the wind speed at the hub position of the upstream wind turbine at time t; u(T,i) is the wind speed at the height of the hub of the downstream wind turbine at time T. Step 4) Considering the effect of the wake slack, calculate the weighting factor for the wake slack correction. ; ； In the formula, This represents the weighting factor for the wake correction. r represents the projected area of ​​the wake region of the i-th upstream wind turbine onto the rotation plane of the j-th downstream wind turbine; j Let be the radius of the rotor of the j-th wind turbine; ； ； In the formula, r x d is the wake radius when the i-th wind turbine reaches the location of the j-th wind turbine; ij r is the distance between the centers of the i-th and j-th wind turbine hubs under vertical wind direction. j It is the radius of the rotor of the j-th downstream wind turbine; This represents the projected area of ​​the wake region of the upstream i-th wind turbine onto the rotation plane of the downstream j-th wind turbine rotor; These are the included angles between the centers of the i-th and j-th wind turbines and the wake influence section, respectively; Step 5) By superimposing the wind speeds arriving at the wind turbine at the same time using a sum-of-squares model, the wind speed at the j-th downstream wind turbine hub position at that time is obtained: ； In the formula, N is the number of upstream wind turbines; n is the nth upstream wind turbine; It is the wake of the nth upstream wind turbine at the jth downstream wind turbine position at time T; It is the wind speed at the hub position of the nth wind turbine upstream at time t; It is the wind speed at the hub height of the j-th downstream wind turbine after superposition at time T; The wind speed at time t is the wind speed measured by the wind tower. It is the weighting factor for the wake correction.

2. The method for reducing wind farm turbine malfunctions based on short-term wind speed forecasting according to claim 1, characterized in that, In step one, the layout information of the entire wind farm is obtained, specifically by: modeling the position of each wind turbine in the wind farm, and finding the relative position of the meteorological tower relative to the entire wind farm; the historical wind speed data recorded by the meteorological tower, including wind speed and wind direction information, and removing unreasonable items in the wind speed information.

3. The method for reducing wind farm turbine malfunctions based on short-term wind speed forecasting according to claim 1, characterized in that, The specific steps of step two are as follows: 1) Set 95% of the historical wind speed data from the wind farm's wind measurement towers as the training dataset and the remaining 5% as the test dataset. Train an LSTM neural network to obtain a prediction model. 2) Use the last data point in the training dataset as input for wind speed prediction, and use the established prediction model to predict wind speed.

4. The method for reducing wind farm unit malfunctions based on short-term wind speed forecasting according to claim 1, characterized in that, In step four: 1) Optimization is achieved through fan start / stop control when the wind speed does not exceed the set minimum threshold. Specifically: If the incoming wind speed remains fluctuating around the cut-in wind speed for a period of time, and does not exceed 120% of the cut-in wind speed, the wind turbine should be shut down through the control system to avoid frequent start-stop of the wind turbine. 2) When the wind speed is higher than the set maximum threshold, the optimization purpose is achieved by variable pitch control; variable pitch adjustment refers to rotating the blades along the longitudinal axis of the blades, and through fine angle changes, making the blades rotate in the feathering direction, changing the blade speed, and realizing the power control of the unit.

5. A processor, characterized in that, The processor is configured to operate according to instructions to execute the method for reducing wind farm unit malfunctions based on short-term wind speed forecasting as described in any one of claims 1-4.

6. A storage medium, characterized in that, The storage medium stores a computer program thereon, the storage medium being used to store instructions for implementing the method for reducing wind farm unit malfunctions based on short-term wind speed forecasting as described in any one of claims 1-4; when the computer program is executed by a processor, it implements the method for reducing wind farm unit malfunctions based on short-term wind speed forecasting as described in any one of claims 1-4.

Images