A wind energy capture method and system that takes into account both efficiency and load at low wind speed

Abstract

The present invention proposes a wind energy capture method and system that takes into account both efficiency and load at low wind speeds, and relates to the technical field of high-power offshore wind power generation. The specific scheme includes: at low wind speeds, based on historical time series data of wind speed, using a long short-term memory neural network to perform real-time wind speed prediction; calculating a reference speed through the predicted value of the wind speed; within a limited speed change rate range of the wind turbine, using a preset time length as a prediction time window, using a genetic algorithm to fit an optimal speed reference curve for each prediction time window, and rolling optimization in turn, and inputting the optimized optimal speed into the wind turbine control system in advance to capture wind energy that takes into account both efficiency and load; the present invention increases the amount of wind energy captured while reducing the system drive shaft load, reduces the wind turbine maintenance cost to a certain extent, and extends the power generation life of the wind turbine.

Description

Technical Field

[0001] The present invention belongs to the technical field of high-power offshore wind power generation, and in particular relates to a wind energy capture method and system that takes both efficiency and load into consideration at low wind speeds. Background Art

[0002] As a renewable energy source, wind energy has the advantages of wide resource distribution, zero carbon emissions and economy. It is regarded as one of the important directions of future energy development. At present, wind power generation plays a vital role in the global renewable energy field. With the increasing penetration rate of wind power grid connection, the randomness and uncertainty of wind power lead to strong volatility in output power, which affects voltage safety, frequency stability and other aspects. Smoothing wind power output can reduce the impact of wind power fluctuations on the power grid. In addition, the unit capacity of wind turbines is getting larger and larger. At present, the mainstream units have reached 10MW, and the newly released offshore wind turbines have reached 20MW, with a blade length of 134 meters. The slow dynamic response characteristics of wind turbines are becoming more and more obvious, and the reduction in wind energy capture efficiency caused by the delayed response of large inertia units under random wind conditions is becoming more and more obvious.

[0003] Therefore, in the traditional wind power generation control system, when the wind speed is less than the rated wind speed of the wind turbine, the wind turbine adopts the maximum wind power capture (MPPT) control strategy. However, as the scale of wind turbines becomes larger and larger, the slow dynamic characteristics of the speed response lag of large inertia units become more and more obvious, and the units are difficult to respond to the rapid changes in turbulent wind speed in time, which leads to a decrease in the efficiency of wind energy captured by the wind turbine. At the same time, due to the randomness and volatility of offshore wind speed changes, wind turbines face serious load problems; the current solutions mainly include low-pass filtering and smoothing the reference speed of the wind turbine, limiting the speed change rate, superimposing constant compensation on the reference speed in the high wind speed area, and limiting the minimum speed. In addition, there is also a strategy of smoothing by adding an energy storage system. When the power generation is higher than the power consumption, the energy storage system stores energy; when the power consumption is higher than the power generation, the energy storage system releases energy, so as to achieve the effect of smoothing power.

[0004] In summary, most existing research is limited to the design and improvement of speed tracking controllers, but this cannot fundamentally solve the above problems and it is difficult to balance load and efficiency. Summary of the invention

[0005] In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a wind energy capture method and system that takes into account both efficiency and load at low wind speeds, thereby increasing the amount of wind energy captured while reducing the system transmission shaft load, reducing the wind turbine maintenance cost to a certain extent, and extending the wind turbine's power generation life.

[0006] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0007] A first aspect of the present invention provides a wind energy capture method that takes into account both efficiency and load at low wind speeds.

[0008] A wind energy capture method that takes into account both efficiency and load at low wind speeds, comprising:

[0009] At low wind speeds, a long short-term memory neural network is used to perform real-time wind speed forecasting based on historical time series data of wind speed;

[0010] Calculate the reference speed through the predicted value of wind speed;

[0011] Under the limited range of wind turbine speed change rate, with the preset time length as a prediction time window, a genetic algorithm is used to fit an optimal speed reference curve for each prediction time window, and the optimal speed obtained by optimization is input into the wind turbine control system in advance to capture wind energy that takes into account both efficiency and load.

[0012] Furthermore, the specific formula for calculating the reference speed is:

[0013]

[0014] in, Optimal tip speed ratio for fan design, is the predicted value of wind speed, is the wind wheel radius.

[0015] Furthermore, the speed change rate range is considered to be When the size of is limited, the maximum and minimum values ​​of the speed change rate are calculated as follows:

[0016]

[0017] in, and are mechanical torque and electromagnetic torque respectively, and are the maximum and minimum values ​​of the electromagnetic torque, is the damping coefficient, is the wind wheel speed.

[0018] Furthermore, the optimal speed reference curve is expressed by a fitting function as follows:

[0019]

[0020] in, The prediction time window The optimal speed, For the A reference speed, The prediction time window The slope of the fitted function in , The prediction time window The intercept of the fitted function in .

[0021] Furthermore, the genetic algorithm is used to fit an optimal speed reference curve for each prediction time window, and different slope-intercept combinations are used as individuals to find the best parameter combination. The specific steps are:

[0022] (1) Create an initial population;

[0023] (2) Calculate the fitness of each individual in the population;

[0024] (3) Selection, crossover, and mutation;

[0025] (4) Calculate the fitness of each individual in the population;

[0026] (5) Determine the termination condition. If the termination condition is met, select the individual with the highest fitness. Otherwise, return to step (3) for iteration.

[0027] Furthermore, the fitness is calculated by minimizing the weighted residual sum of squares, which is expressed as follows:

[0028]

[0029] in, The prediction time window The slope of the fitted function in , The prediction time window The intercept of the fitted function in , For the The weight of the reference speed, is the number of data in the prediction time window, For the A reference speed, For the The fitted value of the optimal speed.

[0030] Furthermore, in the process of finding the best parameter combination, constraints are set according to the speed change rate range. If the fitting result exceeds the constraint condition, the amount of data for fitting in this wind speed interval is gradually reduced until the slope requirement is met by reducing the amount of data.

[0031] A second aspect of the present invention provides a wind energy capture system that takes into account both efficiency and load at low wind speeds.

[0032] A wind energy capture system that takes into account both efficiency and load at low wind speeds includes a wind speed prediction module, a speed calculation module, and a speed optimization module:

[0033] The wind speed prediction module is configured to: perform real-time wind speed prediction based on historical time series data of wind speed using a long short-term memory neural network at low wind speed;

[0034] The rotation speed calculation module is configured to: calculate a reference rotation speed according to a predicted value of the wind speed;

[0035] The speed optimization module is configured as follows: within the speed change rate range of the wind turbine, with a preset time length as a prediction time window, a genetic algorithm is used to fit an optimal speed reference curve for each prediction time window, and the optimization is carried out in a rolling manner. The optimized optimal speed is input into the wind turbine control system in advance to capture wind energy that takes into account both efficiency and load.

[0036] A third aspect of the present invention provides a computer-readable storage medium having a program stored thereon, which, when executed by a processor, implements the steps of a method for capturing wind energy taking into account both efficiency and load at low wind speeds as described in the first aspect of the present invention.

[0037] The fourth aspect of the present invention provides an electronic device, including a memory, a processor, and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps of a method for capturing wind energy that takes into account both efficiency and load at low wind speeds as described in the first aspect of the present invention are implemented.

[0038] One or more of the above technical solutions have the following beneficial effects:

[0039] The present invention aims to optimize the wind energy capture efficiency of the wind turbine and the load borne by the wind turbine transmission shaft, which is specifically embodied in two aspects:

[0040] 1. According to the fan inertia The size of the system is input in advance to obtain the optimal reference speed. , thereby alleviating the slow dynamic characteristics of the wind turbine and improving the efficiency of wind energy capture.

[0041] Second, using genetic algorithms, under the condition of speed constraints, the optimal speed reference curve is fitted, and the drive shaft load of the fan under fluctuating wind is alleviated by limiting the speed change rate to achieve smooth power output.

[0042] Through the above method, the wind turbine can greatly alleviate the load on the wind turbine while ensuring the improvement of wind energy capture efficiency. This is a wind energy capture method that takes both efficiency and load into consideration.

[0043] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0045] Figure 1 This is the topology diagram of the high-power permanent magnet synchronous wind turbine generator system.

[0046] Figure 2 This is the LSTM principle diagram.

[0047] Figure 3 This is the principle diagram of genetic algorithm.

[0048] Figure 4 Optimization flow chart for wind energy capture method considering both efficiency and load. DETAILED DESCRIPTION

[0049] It should be noted that the following detailed descriptions are illustrative and are intended to provide further explanation of the present application. Unless otherwise specified, all technical and scientific terms used in the present invention have the same meanings as those commonly understood by those skilled in the art to which the present application belongs.

[0050] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprise" and / or "include" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0051] The first solution mentioned in the background technology has the disadvantage that the system achieves smooth power output by low-pass filtering the reference speed and limiting the speed change rate, but on this basis, the wind energy capture efficiency is limitedly improved by superimposing constant compensation and limiting the speed change rate. There is no formula reference for the superimposed values. This is obviously not the optimal solution and cannot achieve the maximum capture of wind energy. The second solution of adding an energy storage system can fully capture wind energy and output it smoothly, but the energy storage system itself is expensive and is not suitable for the current actual environment.

[0052] Therefore, the present invention aims to maximize the wind energy capture efficiency without increasing the hardware cost while ensuring the smooth and stable output of wind power. The specific implementation scheme is as follows:

[0053] On the premise of using the long short-term memory neural network to obtain the future wind speed, the corresponding reference speed is calculated, and then the preset time length is used as a prediction time window. Taking into account the speed change rate that the wind turbine can withstand, a genetic algorithm is used to fit an optimal speed reference curve in each prediction time window, and then rolling optimization is performed in sequence, and the optimization results are input into the wind turbine control system; by using this method, the wind energy capture efficiency is improved, the wind turbine output power is smoothed, and the system load is effectively reduced.

[0054] Embodiment 1

[0055] In one embodiment of the present disclosure, a method for capturing wind energy at low wind speed taking into account both efficiency and load is provided, comprising the following steps:

[0056] Step S1: Under low wind speed, based on the historical time series data of wind speed, a long short-term memory neural network is used to perform real-time wind speed prediction.

[0057] Step S2: Calculate the reference rotation speed through the predicted value of wind speed.

[0058] Furthermore, the specific formula for calculating the reference speed is:

[0059]

[0060] in, Optimal tip speed ratio for fan design, is the predicted value of wind speed, is the wind wheel radius.

[0061] Step S3: Within the speed change rate range of the wind turbine, with a preset time length as a prediction time window, use a genetic algorithm to fit an optimal speed reference curve for each prediction time window, and perform rolling optimization in sequence. The optimized optimal speed is input into the wind turbine control system in advance to capture wind energy that takes into account both efficiency and load.

[0062] Furthermore, the speed change rate range is considered to be When the size of is limited, the maximum and minimum values ​​of the speed change rate are calculated as follows:

[0063]

[0064] in, and are mechanical torque and electromagnetic torque respectively, and are the maximum and minimum values ​​of the electromagnetic torque, is the damping coefficient, is the wind wheel speed.

[0065] Furthermore, the optimal speed reference curve is expressed by a fitting function as follows:

[0066]

[0067] in, The prediction time window The optimal speed, For the A reference speed, The prediction time window The slope of the fitted function in , The prediction time window The intercept of the fitted function in .

[0068] Furthermore, the genetic algorithm is used to fit an optimal speed reference curve for each prediction time window, and different slope-intercept combinations are used as individuals to find the best parameter combination. The specific steps are:

[0069] (1) Create an initial population;

[0070] (2) Calculate the fitness of each individual in the population;

[0071] (3) Selection, crossover, and mutation;

[0072] (4) Calculate the fitness of each individual in the population;

[0073] (5) Determine the termination condition. If the termination condition is met, select the individual with the highest fitness. Otherwise, return to step (3) for iteration.

[0074] Furthermore, the fitness is calculated by minimizing the weighted residual sum of squares, which is expressed as follows:

[0075]

[0076] in, The prediction time window The slope of the fitted function in , The prediction time window The intercept of the fitted function in , For the The weight of the reference speed, is the number of data in the prediction time window, For the A reference speed, For the The fitted value of the optimal speed.

[0077] Furthermore, in the process of finding the best parameter combination, constraints are set according to the speed change rate range. If the fitting result exceeds the constraint condition, the amount of data for fitting in this wind speed interval is gradually reduced until the slope requirement is met by reducing the amount of data.

[0078] The scheme proposed in this embodiment first uses the long short-term memory neural network LSTM to predict the wind speed, with 5s as a time prediction window, and then calculates the corresponding reference speed. The genetic algorithm is used to fit the optimal speed of each time window under the condition of limiting the speed change rate to obtain the optimal speed reference. The specific innovative optimization scheme mainly includes two points: one is the advanced reference speed planning based on wind speed prediction to eliminate the slow dynamic response characteristics of the system; the other is the optimal speed reference planning based on the genetic algorithm to smooth the power and increase the wind energy capture efficiency, which is explained in detail below with reference to the figures and formulas.

[0079] 1. Wind power generation system model

[0080] Figure 1 It is the topological diagram of the wind power generation system. First, the topological diagram of the wind power generation system is introduced. The wind power capture of the wind turbine can be expressed as:

[0081]

[0082] in, is the air density, is the rotor radius, is the wind speed, is the wind energy utilization coefficient, wind power It is the energy captured by the wind turbine from the wind per unit time.

[0083] The drive shaft of the fan can be represented by a single mass block model, and its expression is:

[0084]

[0085] in, is the equivalent moment of inertia of the system, and are mechanical torque and electromagnetic torque respectively, is the damping coefficient, is the wind wheel speed, is the rate of change of speed.

[0086] The pitch angle of the wind turbine can be controlled by hydraulic or servo system, which can be expressed as:

[0087]

[0088] in, is the equivalent time constant, and is the pitch angle and its reference value.

[0089] The classic permanent magnet synchronous wind turbine system adopts the predictive control strategy of back-to-back three-level converter. The predictive current control is adopted on both the motor side and the grid side. Since the NPC three-level converter also needs to ensure the DC bus voltage balance, it is necessary to add the DC bus voltage control term. The cost function of the predictive current control of the generator side model can be expressed as:

[0090]

[0091] The cost function of the grid-connected model predictive current control can be expressed as:

[0092]

[0093] in, represents the stator current, Represents the grid current, is the voltage difference between the two capacitors on the DC bus, is the bus capacitor voltage balance weight coefficient.

[0094] For megawatt-level wind turbines, as the blade size becomes larger and larger, the corresponding rotational inertia also becomes larger and larger; therefore, when facing random wind and fluctuating wind, the response lag problem of the wind turbine becomes more and more serious, and the load on the transmission shaft becomes larger and larger, which not only loses a lot of wind energy, but also causes serious damage to the structure of the wind turbine.

[0095] 2. Advance reference speed planning based on wind speed forecast

[0096] First, the long short-term memory neural network LSTM is used to predict the ultra-short-term wind speed. The structure of the long short-term memory neural network LSTM is as follows: Figure 2 As shown in the figure, in order to solve the problem that historical information in recurrent neural networks cannot be transmitted for a long time, LSTM introduces a gating mechanism (Gated Cell) to achieve the ability to learn long time series; LSTM consists of three gates: input gate, forget gate and output gate, which allow LSTM to selectively process and retain relevant information in its memory cells; using LSTM neural network to predict ultra-short-term wind speed has many benefits. Wind speed data usually has time series characteristics, and LSTM can capture the temporal dependencies in the data; at the same time, wind speed data often has complex nonlinear characteristics. LSTM can capture these nonlinear relationships through its multi-layer structure and activation function, thereby improving prediction accuracy.

[0097] The prediction process of LSTM is expressed as follows:

[0098]

[0099]

[0100]

[0101]

[0102]

[0103]

[0104]

[0105] in, , , represents the weight coefficient, , , Indicates deviation, represents the input at time t, that is, the wind speed at time t, and Represents the output at the current moment, and Indicates the cell status, An updated value representing the cell state, , and Represent the forget gate, input gate, and output gate respectively; in these gating mechanisms, the elements in the matrix , and The value is between [0,1]. Represents the activation function.

[0106] After obtaining the predicted wind speed, the reference speed can be obtained by calculation. The formula is as follows:

[0107] (1)

[0108] in, Optimal tip speed ratio for fan design, is the wind speed predicted using the LSTM neural network, is the wind wheel radius.

[0109] 3. Optimal reference speed curve fitting based on genetic algorithm

[0110] When the wind speed changes greatly, the slow dynamic characteristics of the wind turbine result in a limited rate of change of the wind rotor speed; therefore, when considering the electromagnetic torque of the generator When the size of is limited, the range of the wind rotor speed change rate will also be limited by the following formula:

[0111]

[0112] After obtaining the speed change rate range of the wind turbine, the genetic algorithm is used to fit the reference speed with a 5s prediction time window. The specific process is as follows:

[0113] Assume that the number of data in each prediction time window is , then the reference speed is calculated by formula (1):

[0114]

[0115] Since the power formula of the fan is , we can know that the wind power captured by the wind turbine is proportional to the cube of the wind speed, so the wind turbine can capture more wind energy in high wind speed areas. Based on this, when using the genetic algorithm for fitting, the weights of different wind speed areas are considered, and the weight calculation formula is:

[0116]

[0117] in, is the maximum value of the reference speed in the current prediction time window. The function of this formula is to normalize the data and calculate the relative weight of each data point. Larger data points will have higher weights, while smaller data points have relatively lower weights. Therefore, when calculating the weighted residual sum of squares, larger wind speed data can have a greater impact on the fitting results, thereby increasing wind energy capture.

[0118] Assume that the fitting function of each segment is a linear equation:

[0119]

[0120] in, The prediction time window The optimal speed, For the A reference speed, The prediction time window The slope of the fitted function in , The prediction time window The intercept of the fitting function in ; on this basis, the objective function is defined , which is used to minimize the weighted residual sum of squares:

[0121]

[0122] in, The prediction time window The slope of the fitted function in , The prediction time window The intercept of the fitted function in , For the The weight of the reference speed, is the number of data in the prediction time window, For the A reference speed, For the The fitted value of the optimal speed.

[0123] The optimization method is genetic algorithm. The structural principle diagram of genetic algorithm is as follows: Figure 3 As shown in the figure. The genetic algorithm can effectively handle the nonlinear fitting problem of multiple segments of data, avoid falling into the local optimal solution, and thus find the global optimal slope; in addition, the adaptive characteristics of the genetic algorithm enable it to automatically identify the best parameter combination, adapt to changes under different wind speed conditions, and improve the fitting accuracy; by processing multiple candidate solutions in parallel, the genetic algorithm also speeds up the optimization process, ensuring efficient wind energy capture in low wind speed environments, which helps to improve the overall energy conversion efficiency.

[0124] During the optimization process, the constraints need to be set to If the fitting result exceeds this range, the amount of data for fitting in this wind speed interval is gradually reduced, i.e., step-by-step fitting data until the slope requirement can be met by reducing the amount of data. If the amount of data is reduced to 0 and still cannot meet the range requirement, the slope of this interval is directly defined as the boundary value. If the slope requirement is met by reducing the amount of data, the data that is not involved in the fitting in this interval is synchronized to the next time interval to participate in the fitting.

[0125] In order to ensure the continuity of the fitting, the intercept needs to be updated to the end value of the previous segment, and the fitting result is finally output:

[0126]

[0127] The amount of data used in each prediction time window is:

[0128]

[0129] The curve fitted by the above method is the optimal speed reference curve finally input to the fan, that is, Figure 4 In .

[0130] The complete optimization process is as follows Figure 4 As shown, it is necessary to calculate the rotational inertia of the fan The actual size is based on the fan inertia The size of the system is input in advance to obtain the optimal reference speed. , so as to alleviate the slow dynamic characteristics of the wind turbine and improve the wind energy capture efficiency; using genetic algorithms, under the condition of speed constraints, the optimal speed reference curve is fitted, and the transmission shaft load of the wind turbine under fluctuating wind is alleviated by limiting the speed change rate, thereby achieving smooth power output.

[0131] Embodiment 2

[0132] In one embodiment of the present disclosure, a wind energy capture system that takes into account both efficiency and load at low wind speed is provided, including a wind speed prediction module, a rotation speed calculation module, and a rotation speed optimization module:

[0133] The wind speed prediction module is configured to: perform real-time wind speed prediction based on historical time series data of wind speed using a long short-term memory neural network at low wind speed;

[0134] The rotation speed calculation module is configured to: calculate a reference rotation speed according to a predicted value of the wind speed;

[0135] The speed optimization module is configured as follows: within the speed change rate range of the wind turbine, with a preset time length as a prediction time window, a genetic algorithm is used to fit an optimal speed reference curve for each prediction time window, and the optimization is carried out in a rolling manner. The optimized optimal speed is input into the wind turbine control system in advance to capture wind energy that takes into account both efficiency and load.

[0136] Embodiment 3

[0137] The purpose of this embodiment is to provide a computer-readable storage medium.

[0138] A computer-readable storage medium stores a computer program thereon, which, when executed by a processor, implements the steps of a method for capturing wind energy taking into account both efficiency and load at low wind speeds as described in Embodiment 1 of the present disclosure.

[0139] Embodiment 4

[0140] The purpose of this embodiment is to provide an electronic device.

[0141] An electronic device includes a memory, a processor, and a program stored in the memory and executable on the processor. When the processor executes the program, the steps of a wind energy capture method that takes into account both efficiency and load at low wind speed as described in Embodiment 1 of the present disclosure are implemented.

[0142] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

Claims

1. A method for capturing wind energy at low wind speeds taking into account both efficiency and load, characterized in that: include: At low wind speeds, a long short-term memory neural network is used to perform real-time wind speed forecasting based on historical time series data of wind speed; Calculate the reference speed through the predicted value of wind speed; Under the limited speed change rate range of the wind turbine, a preset time length is used as a prediction time window, and a genetic algorithm is used to fit an optimal speed reference curve for each prediction time window. The optimal speed obtained by the optimization is input into the wind turbine control system in advance to capture wind energy that takes into account both efficiency and load. The method of using a genetic algorithm to fit an optimal speed reference curve for each prediction time window is to use different slope-intercept combinations as individuals to find the best parameter combination. The specific steps are as follows: (1) Create an initial population; (2) Calculate the fitness of each individual in the population; (3) Selection, crossover, and mutation; (4) Calculate the fitness of each individual in the population; (5) Determine the termination condition. If the termination condition is met, select the individual with the highest fitness. Otherwise, return to step (3) for iteration.

2. A method for capturing wind energy at low wind speeds taking into account both efficiency and load as claimed in claim 1, characterized in that: The specific formula for calculating the reference speed is: ， in, Optimal tip speed ratio for fan design, is the predicted value of wind speed, is the wind wheel radius.

3. A method for capturing wind energy at low wind speeds taking into account both efficiency and load as claimed in claim 1, characterized in that: The speed change rate range is considered to be the electromagnetic torque of the generator. When the size of is limited, the maximum and minimum values ​​of the speed change rate are calculated as follows: ， in, and are mechanical torque and electromagnetic torque respectively, and are the maximum and minimum values ​​of the electromagnetic torque, is the damping coefficient, is the wind wheel speed.

4. A method for capturing wind energy at low wind speeds taking into account both efficiency and load as claimed in claim 1, characterized in that: The optimal speed reference curve is expressed by a fitting function as follows: ， in, The prediction time window The optimal speed, For the A reference speed, The prediction time window The slope of the fitted function in , The prediction time window The intercept of the fitted function in .

5. A method for capturing wind energy at low wind speeds taking into account both efficiency and load as claimed in claim 1, characterized in that: The fitness is calculated by minimizing the weighted residual sum of squares and is expressed as follows: ， in, The prediction time window The slope of the fitted function in , The prediction time window The intercept of the fitted function in , For the The weight of the reference speed, is the number of data in the prediction time window, For the A reference speed, For the The fitted value of the optimal speed.

6. A method for capturing wind energy at low wind speeds taking into account both efficiency and load as claimed in claim 1, characterized in that: In the process of finding the best parameter combination, set the constraint conditions according to the speed change rate range. If the fitting result exceeds the constraint condition, the amount of data for fitting in this wind speed interval is gradually reduced until the slope requirement is met by reducing the amount of data.

7. A wind energy capture system that takes into account both efficiency and load at low wind speeds, characterized in that: A method for capturing wind energy at low wind speeds taking into account both efficiency and load as described in any one of claims 1 to 6 is implemented, comprising a wind speed prediction module, a speed calculation module and a speed optimization module: The wind speed prediction module is configured to: perform real-time wind speed prediction based on historical time series data of wind speed using a long short-term memory neural network at low wind speed; The rotation speed calculation module is configured to: calculate a reference rotation speed according to a predicted value of the wind speed; The speed optimization module is configured as follows: within the speed change rate range of the wind turbine, with a preset time length as a prediction time window, a genetic algorithm is used to fit an optimal speed reference curve for each prediction time window, and the optimization is carried out in a rolling manner. The optimized optimal speed is input into the wind turbine control system in advance to capture wind energy that takes into account both efficiency and load.

8. An electronic device, comprising: a memory for non-transitory storage of computer readable instructions; a processor for executing the computer-readable instructions; Wherein, when the computer-readable instructions are executed by the processor, the method according to any one of claims 1 to 6 is executed.

9. A storage medium, characterized in that: The computer-readable instructions are non-transitorily stored, wherein when the computer-readable instructions are executed by a computer, the method according to any one of claims 1 to 6 is performed.

Images