Doppler lidar wind measurement method and system for wind turbine array wake management

Abstract

The application provides a wind turbine array wake management method and system based on Doppler laser wind measurement, relates to the field of laser technology, and comprises the following steps: deploying a Doppler laser radar in a wind turbine array, collecting real-time wind speed data, wind direction data and turbulence intensity data in front and back directions of each wind turbine; establishing a wake influence model containing a wake influence coefficient, a decay coefficient and a diffusion coefficient based on the data, and calculating a wake influence range, a loss power and a recovery distance; and accordingly, the wind turbine array is controlled in coordination, the yaw angle or the pitch angle of the upstream wind turbine and the rotating speed of each wind turbine are adjusted, and the power generation efficiency of the wind turbine array is optimized. The application improves the wind speed data collection accuracy of the wind field, enhances the accuracy of the wake influence evaluation, and significantly improves the overall power generation efficiency of the wind field.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of laser, in particular to a method and system for managing wake of wind turbine array based on Doppler laser anemometry. BACKGROUND

[0002] During the operation of wind turbine array, the wake generated by upstream wind turbines will cause the inflow wind speed of downstream wind turbines to decrease and the turbulence intensity to increase, resulting in the decrease of power generation efficiency and the increase of load of wind turbines. Studies have shown that due to the influence of wake, the actual power generation of wind farm is 15%-30% lower than the theoretical expectation. Therefore, it is of great significance to accurately monitor and effectively manage the wake of wind turbines for improving the power generation efficiency of wind farm. At present, the mainstream methods for managing wake include wake prediction based on empirical model, real-time monitoring based on wind measurement data and wake optimization based on control strategy of wind turbines.

[0003] However, the existing wake management technology of wind turbine array has the following shortcomings: first, the traditional wind measurement tower has limited arrangement density, which is difficult to obtain wind field data with high spatial resolution in wind farm, resulting in insufficient accuracy of wake influence evaluation; second, most of the existing wake models are based on static assumption, which fails to fully consider the influence of dynamic factors such as wind speed, turbulence intensity and atmospheric stability on wake characteristics, reducing the accuracy of the model; third, the conventional single-machine control strategy focuses on the optimization of power generation efficiency of single wind turbine, lacking overall consideration of the coupling effect of wake of wind turbine group, and it is difficult to realize the collaborative optimization at the level of wind farm.

[0004] Therefore, the present application provides a method for managing wake of wind turbine array based on Doppler laser anemometry, which realizes accurate monitoring of wake characteristics by deploying laser radar on each wind turbine, establishes a dynamic wake model considering the influence of multiple factors, and designs a collaborative control strategy for wind turbine group based on the model, so as to improve the overall power generation efficiency of wind farm and solve the technical problems of insufficient wind measurement accuracy, low model accuracy and lack of collaborative control in the prior art. SUMMARY

[0005] The embodiments of the present application provide a method and system for managing wake of wind turbine array based on Doppler laser anemometry, which can solve the problems in the prior art.

[0006] The first aspect of the embodiments of the present application is,

[0007] The present application provides a method for managing wake of wind turbine array based on Doppler laser anemometry, which comprises:

[0008] Doppler lidar is deployed in each wind turbine of the wind turbine array to collect real-time wind speed data, wind direction data and turbulence intensity data in front and back directions of each wind turbine in the wind turbine array; the Doppler lidar calculates the real-time wind speed data, wind direction data and turbulence intensity data based on Doppler frequency shift of scattered light of a laser beam reflected by air particles by emitting the laser beam and receiving the scattered light; the Doppler lidar sets a wind sampling point every 20 meters along an axial direction of the laser beam, and a wind measurement distance is 40 meters to 200 meters;

[0009] A wake influence model of the wind turbine array is established according to the real-time wind speed data, wind direction data and turbulence intensity data; the wake influence model includes a wake influence coefficient of an upstream wind turbine on a downstream wind turbine, a wake attenuation coefficient and a wake diffusion coefficient; the wake influence range, the wake loss power and the wake recovery distance of each wind turbine in the wind turbine array are calculated through the wake influence model;

[0010] The wind turbine array is cooperatively controlled based on the wake influence range, the wake loss power and the wake recovery distance; the cooperative control includes: when it is detected that the wake of an upstream wind turbine causes power generation efficiency of a downstream wind turbine to decrease by more than a set threshold, the yaw angle or the pitch angle of the upstream wind turbine is controlled to make the wake area of the upstream wind turbine avoid the downstream wind turbine; and the rotational speed of each wind turbine in the wind turbine array is dynamically adjusted according to the wake influence model to realize overall power generation efficiency optimization of the wind turbine array.

[0011] The Doppler lidar calculates the real-time wind speed data, wind direction data and turbulence intensity data based on Doppler frequency shift of scattered light of a laser beam reflected by air particles by emitting the laser beam and receiving the scattered light includes:

[0012] A narrow-band laser beam with a wavelength of 1.5 μm is emitted by the Doppler lidar, a reference light frequency of the Doppler lidar is offset by 80 MHz relative to a frequency of the narrow-band laser beam, and scattered light signals of the narrow-band laser beam scattered by air particles are collected; the scattered light signals and the reference light are heterodyne mixed on a photodetector to obtain beat signals;

[0013] A velocity azimuth angle scanning scheme is adopted, the laser beam of the Doppler lidar is controlled to rotate and scan in a horizontal plane at an elevation angle of 30 degrees, the scattered light signals are collected every 30 degrees of azimuth angle, and the beat signals at 12 azimuth angles are obtained in a range of 360 degrees; the beat signals are subjected to fast Fourier transform to obtain Doppler spectrum, and radial wind speed is calculated based on a peak position of the Doppler spectrum;

[0014] Adaptive threshold denoising is performed on the Doppler spectrum by using wavelet transform, and reliable data is screened by monitoring signal quality indicators in real time; more than 100 valid samples are accumulated for each measuring point, a horizontal wind speed vector is reconstructed by least square fitting to obtain wind speed and wind direction angle; a ratio of standard deviation to average value of the radial wind speed is calculated to obtain turbulence intensity; the time resolution of the wind field parameter solving method is 1 second, and the spatial resolution is 20 meters.

[0015] According to the real-time wind speed data, wind direction data and turbulence intensity data, a wake influence model of the wind turbine array is established; the wake influence model includes a wake influence coefficient of an upstream wind turbine on a downstream wind turbine, a wake attenuation coefficient and a wake diffusion coefficient, and includes:

[0016] Real-time wind speed data of a downstream area of the wind turbine is collected, and a speed loss Gaussian distribution function is established based on the real-time wind speed data:

[0017]

[0018] Wherein V(x, r) represents the wind speed at a downstream distance x and a radial distance r, V∞ is the incoming flow wind speed, CT is the thrust coefficient, α(x) is the wake influence coefficient, and σ(x) is the wake diffusion width;

[0019] Wind farm turbulence intensity data is obtained, and a wake attenuation coefficient is calculated based on the turbulence intensity data:

[0020] β=0.35+0.65TJ,

[0021] Wherein β is the wake attenuation coefficient, and TI is the turbulence intensity;

[0022] The wake attenuation coefficient β is substituted into the wake influence coefficient calculation formula:

[0023]

[0024] The wake influence coefficient α(x) is obtained;

[0025] Wherein x is the downstream distance, and D is the rotor diameter;

[0026] A wake diffusion model considering wind direction deviation is established based on wind direction data:

[0027]

[0028] Wherein k is the wake diffusion coefficient, and the wake diffusion model is substituted into the speed loss Gaussian distribution function;

[0029] For the case that a downstream wind turbine is simultaneously affected by the wakes of multiple upstream wind turbines, a dynamic wake model is established based on the dynamic wake model of the National Renewable Energy Laboratory (NREL) and the wind direction data:

[0030] The principle of mass conservation is used to establish a wake superposition model:

[0031]

[0032] where ΔV total is the superposed velocity deficit, and ΔV i is the velocity deficit caused by a single wake;

[0033] The laser radar wind speed measurement value V measured is obtained in real time, and the laser radar wind speed measurement value is compared with the model predicted wind speed value V model to calculate a parameter correction amount:

[0034] δp=γ(V measureed -V model ),

[0035] where δp is the parameter correction amount, and γ is a learning rate;

[0036] The wake influence coefficient, the wake attenuation coefficient, and the wake diffusion coefficient are adaptively corrected based on the parameter correction amount.

[0037] The wake influence range, the wake loss power, and the wake recovery distance of each wind turbine in the wind turbine array are calculated by the wake influence model, and the wake influence range, the wake loss power, and the wake recovery distance include:

[0038] A cylindrical coordinate system is established with the center of the wind wheel as the origin and the incoming flow direction as the axial direction, an initial calculation grid is constructed in the cylindrical coordinate system, a grid encryption coefficient is determined by calculating the velocity gradient, the calculation grid is adaptively encrypted based on the grid encryption coefficient, and an optimized calculation grid is obtained;

[0039] At each grid point of the optimized calculation grid, a velocity deficit value of a position point downstream of the wind turbine is calculated using a velocity deficit Gaussian distribution function, and the ratio of the velocity deficit value to the incoming flow speed is determined as a relative velocity deficit;

[0040] Based on the relative velocity deficit, by scanning the grid points in the calculation grid, a space region with a relative velocity deficit greater than a first preset threshold value is determined as the wake influence range of the wind turbine;

[0041] Within the wake influence range, a power curve model of the wind turbine is established, a wake loss power is calculated based on the power curve model and the velocity deficit value, and a power loss distribution of the wind turbine within the wake influence range is obtained;

[0042] Calculate a wind speed recovery rate along a center line of the wind wheel based on the wake influence range, and determine a downstream distance corresponding to the wind speed recovery rate greater than a second preset threshold as a wake recovery distance, where the wind speed recovery rate is a ratio of a wind speed at any position on the center line to an incoming flow wind speed;

[0043] Store time sequence data of the wake influence range, the power loss distribution, and the wake recovery distance into a wind turbine operation characteristic database for cooperative control of the wind turbine.

[0044] The cooperative control includes: when it is detected that a wake of an upstream wind turbine causes a power generation efficiency of a downstream wind turbine to decrease by more than a set threshold, controlling a yaw angle or a pitch angle of the upstream wind turbine so that a wake area of the upstream wind turbine avoids the downstream wind turbine, including:

[0045] Real-time measurement of a three-dimensional wind speed distribution in a wake area of an upstream wind turbine by a laser radar system to obtain wake characteristic data of the upstream wind turbine;

[0046] Collection of actual power output data of a downstream wind turbine by a wind farm monitoring system, comparison of the actual power output data with theoretical power output data, and calculation of a power generation efficiency loss of the downstream wind turbine;

[0047] When the power generation efficiency loss exceeds a preset loss threshold, calculation of an optimal yaw angle of the upstream wind turbine based on the wake characteristic data and environmental parameters by a dynamic programming algorithm;

[0048] Yaw adjustment of the upstream wind turbine according to the optimal yaw angle, and calculation of an optimal pitch angle of the upstream wind turbine when the power generation efficiency loss after the yaw adjustment still exceeds the preset loss threshold;

[0049] Coordination and adjustment of the yaw angle and the pitch angle of the upstream wind turbine by a fuzzy PID algorithm until the power generation efficiency loss of the downstream wind turbine is lower than the preset loss threshold;

[0050] Real-time monitoring of a power generation efficiency recovery degree of the downstream wind turbine, a power generation loss of the upstream wind turbine, and a device load condition, and online optimization and adjustment of control parameters based on the monitoring results.

[0051] Dynamic adjustment of rotational speeds of wind turbines in the wind turbine array according to the wake influence model to achieve optimalization of overall power generation efficiency of the wind turbine array, including:

[0052] The three-dimensional wind field data of the wind farm is collected by a wind measurement tower and a laser radar system, the three-dimensional wind field data is input into a Jensen wake model, and a dynamic wake characteristic model considering wind speed, turbulence intensity and atmospheric stability is established;

[0053] A wake influence coefficient matrix between wind turbines in the wind turbine array is calculated based on the dynamic wake characteristic model, and the wake influence coefficient matrix represents the wake interaction strength between any two wind turbines;

[0054] The wake influence coefficient matrix is input into a wind turbine array power output model, a mapping relationship between the rotational speed of each wind turbine and the total power generation of the wind farm is established, and the mapping relationship considers the influence of the rotational speed on the power generation efficiency of a single machine and the wake characteristics;

[0055] A multi-objective optimization problem is constructed based on the mapping relationship, an improved particle swarm algorithm with an adaptive weight factor is used to solve the multi-objective optimization problem, and an optimal rotational speed distribution scheme for each wind turbine is obtained;

[0056] The wind turbine array is divided into multiple control sub-regions, the optimal rotational speed distribution scheme is executed in each control sub-region, and the operating state data of each wind turbine is collected in real time through a SCADA system;

[0057] The effect of rotational speed control is evaluated based on the operating state data, when a deviation between the actual power generation efficiency and the expected efficiency is detected, the parameters of the dynamic wake characteristic model are corrected online, and the optimal rotational speed distribution scheme is recalculated.

[0058] A multi-objective optimization problem is constructed based on the mapping relationship, an improved particle swarm algorithm with an adaptive weight factor is used to solve the multi-objective optimization problem, and an optimal rotational speed distribution scheme for each wind turbine is obtained including:

[0059] A multi-objective optimization model of the wind turbine array is established, the multi-objective optimization model includes a total power generation of the wind farm objective function, a wake interference objective function and a unit load objective function; wherein the total power generation of the wind farm objective function is the sum of the output power of each wind turbine, the wake interference objective function is the product of the wake influence coefficient between each wind turbine and the speed loss value, and the unit load objective function is the maximum load value of each wind turbine;

[0060] A particle coding scheme of an improved particle swarm algorithm is constructed based on the multi-objective optimization model, the rotational speed of each wind turbine is used as an optimization variable, the particle swarm is initialized, and the objective function values corresponding to each particle are calculated;

[0061] An inertia weight of the improved particle swarm algorithm is calculated by using a nonlinear decreasing strategy, the inertia weight is calculated by a power function of a ratio of a current iteration number to a maximum iteration number, and an index of the power function is a nonlinear adjustment factor;

[0062] A distance between each particle and a global optimal solution is calculated, and individual learning factors and social learning factors are dynamically calculated by an exponential function based on the distance; the individual learning factors are increased to strengthen a local search ability when the distance is large, and the social learning factors are increased to accelerate a convergence speed when the distance is small;

[0063] A speed and a position of a particle are updated according to the inertia weight, the individual learning factors and the social learning factors, and a target function value of the updated particle is calculated;

[0064] A rank of each particle is obtained by non-dominated sorting of the updated particle swarm, and a crowding distance of the particle is calculated in the same rank; and a high-quality particle is selected to form a new generation population based on the rank and the crowding distance;

[0065] It is judged whether a target function improvement rate of the new generation population, when the target function improvement rate of a preset number of consecutive generations is lower than a set threshold value or reaches a maximum iteration number, a rotating speed configuration scheme corresponding to a global optimal particle is taken as an optimal rotating speed distribution scheme of each wind turbine generator unit; if the convergence condition is not met, the step of calculating the inertia weight is returned to continue iterative optimization.

[0066] A second aspect of the embodiment of the application,

[0067] A wind turbine generator array wake management system based on Doppler laser wind measurement is provided, comprising:

[0068] A first unit is configured to deploy a Doppler laser radar in each wind turbine generator of a wind turbine generator array, and collect real-time wind speed data, wind direction data and turbulence intensity data in front and back directions of each wind turbine generator in the wind turbine generator array; the Doppler laser radar calculates the real-time wind speed data, wind direction data and turbulence intensity data based on Doppler frequency shift of scattered light after reflection of a laser beam by air particles; the Doppler laser radar sets a wind measurement sampling point every 20 meters along an axial direction of the laser beam, and a wind measurement distance is 40-200 meters;

[0069] A second unit is configured to establish a wake influence model of the wind turbine generator array according to the real-time wind speed data, wind direction data and turbulence intensity data; the wake influence model comprises a wake influence coefficient, a wake attenuation coefficient and a wake diffusion coefficient of an upstream wind turbine generator on a downstream wind turbine generator; the wake influence model is used to calculate a wake influence range, a wake loss power and a wake recovery distance of each wind turbine generator in the wind turbine generator array.

[0070] a third unit configured to perform cooperative control on the wind turbine array based on the wake influence range, the wake loss power and the wake recovery distance; the cooperative control comprises: when it is detected that the wake of an upstream wind turbine causes the power generation efficiency of a downstream wind turbine to decrease by more than a set threshold, controlling the yaw angle or the pitch angle of the upstream wind turbine to make the wake area of the upstream wind turbine avoid the downstream wind turbine; and dynamically adjusting the rotational speed of each wind turbine in the wind turbine array according to the wake influence model to achieve the optimization of the overall power generation efficiency of the wind turbine array.

[0071] a third aspect of the embodiment of the application,

[0072] An electronic device is provided, comprising:

[0073] a processor;

[0074] a memory for storing processor-executable instructions;

[0075] wherein the processor is configured to invoke the instructions stored in the memory to perform the method described above.

[0076] a fourth aspect of the embodiment of the application,

[0077] A computer-readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method described above.

[0078] The beneficial effects of the present application are as follows:

[0079] 1. Improve measurement accuracy and real-time performance: By deploying Doppler laser radars on each wind turbine and obtaining real-time wind speed, wind direction and turbulence intensity data within a range of 40-200 meters at a high spatial resolution (one sampling point every 20 meters), the complex wind condition changes within the wind turbine array can be captured more accurately, providing a more reliable data basis for the establishment of the wake model. Compared with traditional wind measurement methods, the dynamic changes of the wake can be reflected more timely and accurately.

[0080] 2. Optimize the wake model and accurately predict the wake influence: The wake influence model established by this method includes key parameters such as wake influence coefficient, wake attenuation coefficient and wake diffusion coefficient, which can more accurately predict the wake influence range, wake loss power and wake recovery distance of the upstream wind turbine on the downstream wind turbine, providing more accurate guidance for subsequent cooperative control strategies.

[0081] 3. Improve the overall power generation efficiency of the wind farm: Based on the accurate wake prediction results, by controlling the yaw angle or pitch angle of the upstream wind turbine and dynamically adjusting the speed of each wind turbine, the influence of the wake on the downstream wind turbine can be effectively reduced, thereby minimizing the power loss caused by the wake and significantly improving the power generation efficiency of the entire wind turbine array. BRIEF DESCRIPTION OF DRAWINGS

[0082] Figure 1 A flowchart of the wind turbine array wake management method of the Doppler laser wind measurement embodiment of the present application is shown in

[0083] Figure 2 A structural diagram of the wind turbine array wake management system of the Doppler laser wind measurement embodiment of the present application is shown in DETAILED DESCRIPTION

[0084] To make the objectives, technical solutions, and advantages of the embodiments of the present application clearer, the technical solutions in the embodiments of the present application will be described below in connection with the drawings of the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments of the present application. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative work fall within the scope of protection of the present application.

[0085] The technical solutions of the present application will be described in detail below with specific embodiments. The following specific embodiments can be combined with each other, and some embodiments may not be described again for the same or similar concepts or processes.

[0086] Figure 1 A flowchart of the wind turbine array wake management method of the Doppler laser wind measurement embodiment of the present application is shown in Figure 1 The method comprises:

[0087] S11. Deploy a Doppler laser radar in each wind turbine of the wind turbine array to collect real-time wind speed data, wind direction data, and turbulence intensity data in front and back directions of each wind turbine in the wind turbine array; the Doppler laser radar calculates the real-time wind speed data, wind direction data, and turbulence intensity data based on the Doppler frequency shift of the scattered light after the laser beam is reflected by air particles by emitting a laser beam and receiving the scattered light; the Doppler laser radar sets a wind measurement sampling point every 20 meters along the axis of the laser beam, and the wind measurement distance is 40-200 meters;

[0088] S12. According to the real-time wind speed data, wind direction data and turbulence intensity data, a wake influence model of the wind turbine array is established; the wake influence model includes a wake influence coefficient of an upstream wind turbine on a downstream wind turbine, a wake attenuation coefficient and a wake diffusion coefficient; a wake influence range, a wake loss power and a wake recovery distance of each wind turbine in the wind turbine array are calculated through the wake influence model;

[0089] S13. Based on the wake influence range, the wake loss power and the wake recovery distance, the wind turbine array is cooperatively controlled; the cooperative control includes: when it is detected that the wake of an upstream wind turbine causes the power generation efficiency of a downstream wind turbine to decrease by more than a set threshold, the yaw angle or the pitch angle of the upstream wind turbine is controlled to make the wake area of the upstream wind turbine avoid the downstream wind turbine; and the rotational speed of each wind turbine in the wind turbine array is dynamically adjusted according to the wake influence model to achieve the optimization of the overall power generation efficiency of the wind turbine array.

[0090] The method for managing the wake of a wind turbine array by Doppler laser wind measurement specifically includes the following steps:

[0091] First, Doppler laser radars are installed on each wind turbine in the wind turbine array. These radars work by emitting a laser beam and receiving scattered light reflected by particles in the air. By analyzing the Doppler shift of the scattered light, real-time wind speed, wind direction and turbulence intensity data can be calculated. To obtain more comprehensive wind field information, each Doppler laser radar sets a wind sampling point every 20 meters from 40 meters to 200 meters along the direction of the laser beam, for example, at 40 meters, 60 meters, 80 meters, and so on, until 200 meters.

[0092] Next, using the collected real-time wind speed, wind direction and turbulence intensity data, a wake influence model of the wind turbine array is established. This model includes three key parameters: a wake influence coefficient, which quantifies the influence of an upstream wind turbine on a downstream wind turbine; a wake attenuation coefficient, which describes the degree to which the wake intensity decreases with distance; and a wake diffusion coefficient, which describes the degree to which the wake spreads laterally as it travels downstream. For example, suppose the wake influence coefficient of upstream wind turbine A on downstream wind turbine B is 0.2, which means that the wind speed at B's location will be reduced by 20% due to A's wake influence. Through this wake influence model, the wake influence range of each wind turbine, which is the area where the wake speed is lower than the undisturbed wind speed by a certain percentage, the wake loss power, which is the power value of the power generation reduction of the downstream wind turbine due to the wake effect, and the wake recovery distance, which is the distance at which the wake speed recovers to near the undisturbed wind speed, can be calculated.

[0093] Then, based on the calculated wake impact range, wake power loss, and wake recovery distance, the wind turbine array is coordinated to improve overall power generation efficiency. The core of the control strategy is this: when the system detects that the wake of an upstream wind turbine has reduced the power generation efficiency of the downstream wind turbine by more than a preset threshold (for example, a 5% reduction in power generation), the upstream wind turbine is controlled. This control can be achieved by adjusting the yaw angle of the upstream wind turbine to deviate from the downstream wind turbine, or by adjusting the pitch angle to change the angle between the blades and the wind, thereby reducing the wake intensity. Simultaneously, the system dynamically adjusts the speed of each wind turbine in the wind turbine array based on the wake impact model to achieve the goal of optimizing overall power generation efficiency. For example, if it is detected that the wake of wind turbine A is having an excessive impact on wind turbine B, the yaw angle of A can be adjusted by 3 degrees to deviate the wake direction from that of B, while the speed of B can be slightly increased to minimize the power loss caused by the wake.

[0094] The beneficial effects of this method can be summarized in the following three aspects:

[0095] 1. Improve the overall power generation efficiency of the wind farm: Through real-time monitoring and dynamic adjustment, the wake loss is minimized, thereby increasing the output power of the entire wind farm.

[0096] 2. Extend the service life of wind turbines: Through coordinated control, the fatigue load of wind turbines in the wake environment is reduced, thereby extending their service life.

[0097] 3. Reduce wind farm operating costs: Higher power generation efficiency and longer equipment life can reduce wind farm operation and maintenance costs and improve economic benefits.

[0098] In an optional embodiment, the Doppler laser radar emits a laser beam and receives scattered light after the laser beam is reflected by air particles, and calculates the real-time wind speed data, wind direction data, and turbulence intensity data based on the Doppler frequency shift of the scattered light, including:

[0099] A narrowband laser beam with a wavelength of 1.5 μm is emitted by a Doppler laser radar, wherein the reference light frequency of the Doppler laser radar is offset by 80 MHz relative to the frequency of the narrowband laser beam, and a scattered light signal of the narrowband laser beam after being scattered by air particles is collected; the scattered light signal is heterodyned with the reference light on a photodetector to obtain a beat frequency signal;

[0100] The Doppler lidar is controlled to rotate and scan a laser beam in a horizontal plane at an elevation angle of 30 degrees, and the scattered light signal is collected every 30 degrees of azimuth angle, and 12 azimuth angles of the beat frequency signal are obtained within a range of 360 degrees; the beat frequency signal is subjected to fast Fourier transform to obtain a Doppler spectrum, and a radial wind speed is calculated based on a peak position of the Doppler spectrum;

[0101] The Doppler spectrum is subjected to adaptive threshold denoising by using a wavelet transform, and reliable data is screened in real time by monitoring a signal quality index; more than 100 effective samples are accumulated for each measurement point, a horizontal wind speed vector is reconstructed by least square fitting, and a wind speed size and a wind direction angle are obtained; a ratio of a standard deviation to an average value of the radial wind speed is calculated to obtain a turbulence intensity; a time resolution of the wind field parameter solving method is 1 second, and a spatial resolution is 20 meters.

[0102] One specific embodiment of the Doppler lidar wind field measurement method is as follows:

[0103] Firstly, a Doppler lidar with a narrow-band pulsed laser with a wavelength of 1.5 μm is used for measurement. The reference light frequency of the laser is offset by 80 MHz relative to the emitted light frequency. The narrow-band laser beam emitted by the laser propagates in the atmosphere and scatters after encountering airborne particulate matter (such as aerosols, dust, water droplets, etc.).

[0104] Next, a receiving unit collects the scattered light signal. The scattered light signal is mixed with the reference light branched from the laser on a photodetector. The result of the heterodyne mixing is to produce a signal with a frequency difference between the two light frequencies, called the beat frequency signal. Since the Doppler shift of the scattered light is proportional to the movement speed of the air particles, the frequency of the beat frequency signal contains the movement information of the air particles, and further reflects the wind speed information.

[0105] In order to obtain three-dimensional wind field information, a velocity azimuth scanning scheme is adopted. The laser beam is controlled to rotate and scan in a horizontal plane at an elevation angle of 30 degrees. The scattered light signal is collected every 30 degrees of azimuth angle, and 12 azimuth angles of the beat frequency signal are obtained within a range of 360 degrees. For example, starting from the north direction, the beat frequency signals of north, north-east 30 degrees, north-east 60 degrees,..., and north-west 30 degrees are collected in turn.

[0106] The beat frequency signal collected at each azimuth angle is subjected to fast Fourier transform to convert it from a time domain signal to a frequency domain signal, and a Doppler spectrum is obtained. The peak position of the Doppler spectrum corresponds to the radial wind speed, i.e. the wind speed component in the direction of the laser beam. For example, if the peak position of the Doppler spectrum corresponds to a frequency offset of 1 MHz, the radial wind speed at that azimuth angle can be calculated according to the corresponding relationship between the Doppler shift and the radial wind speed.

[0107] To improve data quality, wavelet transform is used for adaptive threshold denoising of Doppler spectrum. The threshold is adaptively adjusted according to the noise level of the Doppler spectrum, and the noise signal is filtered out and the effective signal is retained. At the same time, real-time monitoring of signal quality indicators such as signal-to-noise ratio, spectral width, etc. is carried out to screen reliable data. For example, set the signal-to-noise ratio threshold to 10dB, and the data below this threshold will be considered invalid data.

[0108] More than 100 valid samples are accumulated for each measurement point, for example, 100 valid radial wind speed samples are continuously collected at each azimuth angle. Then, the least squares fitting method is used to reconstruct the horizontal wind speed vector. Project the radial wind speed of 12 azimuth angles onto the horizontal plane, and solve the east and north components of the horizontal wind speed by least squares fitting, and then get the wind speed and wind direction angle. For example, through least squares fitting, the east wind speed component is 3m / s, and the north wind speed component is 4m / s, then the wind speed is 5m / s, and the wind direction angle is north by east 53.1 degrees.

[0109] Finally, the ratio of the standard deviation to the average value of the radial wind speed is calculated to obtain the turbulence intensity. For example, if the standard deviation of 100 radial wind speed samples is 0.5m / s and the average value is 5m / s, then the turbulence intensity is 0.1. The time resolution of the wind field parameter solving method is 1 second and the spatial resolution is 20 meters.

[0110] The beneficial effects of this method can be summarized as follows:

[0111] 1. High precision: Using heterodyne mixing technology and narrowband laser, combined with wavelet transform denoising processing, can effectively improve the accuracy and reliability of wind speed measurement.

[0112] 2. High temporal and spatial resolution: 1 second time resolution and 20 meter spatial resolution can finely capture the rapid changes and spatial distribution characteristics of the wind field, providing more detailed wind field information.

[0113] 3. Full-azimuth measurement: The velocity azimuth angle scanning scheme can realize 360-degree full-azimuth wind field measurement and obtain complete wind speed and wind direction information, avoiding measurement blind area.

[0114] In an optional embodiment, a wake influence model of the wind turbine array is established according to the real-time wind speed data, wind direction data and turbulence intensity data; the wake influence model includes the wake influence coefficient of the upstream wind turbine on the downstream wind turbine, the wake attenuation coefficient and the wake diffusion coefficient, including:

[0115] Collect real-time wind speed data in the downstream area of the wind turbine, and establish a velocity deficit Gaussian distribution function based on the real-time wind speed data:

[0116]

[0117] wherein V(x, r) represents the wind speed at downstream distance x, radial distance r, V∞ is the incoming wind speed, CT is the thrust coefficient, a(x) is the wake influence coefficient, and σ(x) is the wake diffusion width;

[0118] obtaining wind farm turbulence intensity data, calculating a wake decay coefficient based on the turbulence intensity data:

[0119] β = 0.35 + 0.65TI,

[0120] wherein β is the wake decay coefficient and TI is the turbulence intensity;

[0121] substituting the wake decay coefficient β into a wake influence coefficient calculation formula:

[0122]

[0123] obtaining a wake influence coefficient a(x);

[0124] wherein x is the downstream distance and D is the wind wheel diameter;

[0125] establishing a wake diffusion model considering wind direction deviation based on wind direction data:

[0126]

[0127] wherein k is the wake diffusion coefficient, substituting the wake diffusion model into the velocity deficit Gaussian distribution function;

[0128] for the case that a downstream wind turbine is simultaneously affected by multiple upstream wind turbines, a wake superposition model is established based on the dynamic

[0129] conservation principle:

[0130]

[0131] wherein ΔV total is the superposed velocity deficit, and ΔV i is the velocity deficit caused by a single wake;

[0132] obtaining a laser radar wind speed measurement value V measured in real time, comparing the laser radar wind speed measurement value with a model predicted wind speed value V model , and calculating a parameter correction amount:

[0133] δp = γ(V measured -V model ),

[0134] wherein δp is the parameter correction amount and γ is the learning rate;

[0135] The wake influence coefficient, the wake decay coefficient, and the wake diffusion coefficient are adaptively modified based on the parameter correction amount.

[0136] First, a real-time monitoring system for a wind turbine array is set up. This system includes multiple sensors, such as anemometers and wind vanes installed on each wind turbine, as well as a laser radar that can measure wind speed and turbulence intensity at different locations within the wind farm. The system transmits all sensor data in real time to a central server for processing.

[0137] Next, the collected data is preprocessed. This includes data cleaning, such as removing outliers and noise, and data synchronization to ensure that all data corresponds to the same time point. For example, if there is an abnormal jump in the data of a certain anemometer, interpolation or smoothing can be used with the data of adjacent time points.

[0138] Then, a velocity deficit Gaussian distribution function model is established based on the preprocessed wind speed data. This model describes the influence of the wake of an upstream wind turbine on the downstream wind speed. Specifically, the model calculates the wind speed at a downstream location, which is related to the incoming wind speed, the thrust coefficient of the wind turbine, the wake influence coefficient, and the wake diffusion width. For example, assuming the incoming wind speed is 10 meters per second, the thrust coefficient is 0.8, the wake influence coefficient is 0.5, and the wake diffusion width is 20 meters, the wind speed at a specific downstream location can be calculated.

[0139] Next, the wake decay coefficient is calculated based on the turbulence intensity data of the wind farm. The wake decay coefficient reflects the degree to which the wake intensity decreases with increasing distance downstream. The higher the turbulence intensity, the faster the wake decays. For example, if the turbulence intensity is 0.1, the corresponding wake decay coefficient can be calculated.

[0140] The wake influence coefficient is obtained by substituting the calculated wake decay coefficient into the wake influence coefficient calculation formula. The wake influence coefficient represents the degree of influence of the upstream wind turbine on the downstream wind speed. For example, based on the wake decay coefficient and the rotor diameter, the wake influence coefficient at a specific downstream distance can be calculated.

[0141] At the same time, a wake diffusion model considering wind direction deviation is established based on wind direction data. This model takes into account the influence of wind direction changes on the shape of the wake. The wake diffusion coefficient determines the diffusion speed of the wake. For example, changes in wind direction will cause the wake to no longer be a simple circle, but an ellipse or other shape.

[0142] The wake diffusion model is substituted into the velocity deficit Gaussian distribution function to obtain more accurate wind speed prediction values.

[0143] For the case where a downstream wind turbine is simultaneously affected by multiple upstream wind turbines' wake, a wake superposition model based on the principle of momentum conservation is used. This model superimposes the effects of multiple wakes to calculate the total speed loss of the downstream wind turbine. For example, if the downstream wind turbine is affected by two upstream wind turbines, the speed losses caused by the two wakes are superimposed to obtain the total speed loss.

[0144] To further improve the accuracy of the model, the laser radar measurement value is used to correct the model. The wind speed value measured by the laser radar is compared with the wind speed value predicted by the model, and the parameter correction amount is calculated. For example, if the wind speed measured by the laser radar is 8 meters per second, and the wind speed predicted by the model is 7 meters per second, the parameter correction amount is 1 meter per second.

[0145] Finally, the wake influence coefficient, wake attenuation coefficient and wake diffusion coefficient are adaptively modified according to the parameter correction amount. This enables the model to adjust according to the actual situation, improving the prediction accuracy. For example, if the parameter correction amount is positive, the values of the wake influence coefficient, wake attenuation coefficient and wake diffusion coefficient are increased, and vice versa.

[0146] The beneficial effects of this method are reflected in three aspects:

[0147] 1. Improve the efficiency of wind power generation: By accurately predicting wind speed, the operation strategy of wind turbines can be optimized, the efficiency of wind energy capture can be improved, and the overall power generation efficiency of the wind farm can be improved.

[0148] 2. Reduce the fatigue load of wind turbines: By accurately predicting wind speed, wind turbines can be prevented from operating in unstable wind conditions, thereby reducing the fatigue load of wind turbines and prolonging their service life.

[0149] 3. Optimize the layout of wind farms: By accurately modeling the wake effect, the arrangement of wind turbines can be optimized to minimize wake loss and improve the overall performance of the wind farm.

[0150] In an alternative embodiment, the wake influence range, wake loss power and wake recovery distance of each wind turbine in the wind turbine array are calculated by the wake influence model, comprising:

[0151] A cylindrical coordinate system with the center of the wind wheel as the origin and the incoming flow direction as the axial direction is established, an initial calculation grid is constructed in the cylindrical coordinate system, a grid encryption coefficient is determined by calculating the velocity gradient, and the calculation grid is adaptively encrypted based on the grid encryption coefficient to obtain an optimized calculation grid.

[0152] At each grid point of the optimized calculation grid, a velocity loss Gaussian distribution function is used to calculate a velocity loss value of a downstream location of the wind turbine, and a ratio of the velocity loss value to the incoming wind speed is determined as a relative velocity loss;

[0153] Based on the relative speed loss, by scanning the grid points in the calculation grid, a spatial region where the relative speed loss is greater than a first preset threshold is determined as a wake influence range of the wind turbine;

[0154] Establishing a power curve model of the wind turbine within the wake influence range, calculating the wake loss power based on the power curve model and the speed loss value, and obtaining a power loss distribution of the wind turbine within the wake influence range;

[0155] Based on the wake influence range, a wind speed recovery rate is calculated along the centerline of the wind rotor, and a downstream distance corresponding to when the wind speed recovery rate is greater than a second preset threshold is determined as a wake recovery distance, wherein the wind speed recovery rate is the ratio of the wind speed at any position on the centerline to the incoming wind speed;

[0156] The time series data of the wake influence range, the power loss distribution and the wake recovery distance are stored in a wind turbine operation characteristic database for use in the coordinated control of the wind turbine.

[0157] First, obtain the layout information of the wind turbine array, including the location coordinates, hub height, blade radius and other parameters of each wind turbine. At the same time, obtain real-time meteorological data, including incoming wind speed, wind direction, air density, etc. Assume that a wind farm has 3 wind turbines with coordinates of (0,0), (5D,0), (10D,0), where D is the rotor diameter, which is assumed to be 150 meters. The incoming wind speed is 10m / s, the wind direction is due east, and the air density is 1.225kg / m 3 .

[0158] Establish a cylindrical coordinate system with the hub center of each wind turbine as the origin, with the positive x-axis aligned with the incoming wind direction. Construct an initial cylindrical computational grid around the turbine. The initial grid can be set to an even spacing, for example, with grid points every 0.5D in the radial direction, every D in the axial direction, and every 6 degrees in the angular direction.

[0159] Next, the velocity gradient at each grid point is calculated. The velocity gradient can be approximated by the difference in velocity between adjacent grid points divided by the grid spacing. Based on the calculated velocity gradient, a grid refinement factor is determined. For example, a threshold value can be set, and when the velocity gradient is greater than the threshold value, the grid in that region is refined. Suppose the velocity gradient threshold is 0.1 s-1, and at 2D downstream of the first wind turbine, the calculated velocity gradient is 0.12 s-1, which is greater than the threshold, so the grid in this region is refined.

[0160] The computational grid is adaptively refined based on the grid refinement factor. For example, the grid spacing in the region with a larger velocity gradient can be reduced by half. Suppose at 2D downstream of the first wind turbine, both the radial and axial grid spacings are reduced to half of the original.

[0161] At each grid point of the optimized computational grid, the velocity deficit value at the downstream location of the wind turbine is calculated using a velocity deficit Gaussian distribution function. The parameters of the velocity deficit Gaussian distribution function can be determined based on factors such as the characteristics of the wind turbine and the incoming wind speed. Suppose at 2D downstream of the first wind turbine, the calculated velocity deficit value is 2 m / s.

[0162] The ratio of the velocity deficit value calculated at each grid point to the incoming wind speed is determined as the relative velocity deficit. For example, at 2D downstream of the first wind turbine, the relative velocity deficit is 2 / 10 = 0.2.

[0163] By scanning all grid points in the computational grid, the spatial region with a relative velocity deficit greater than a first preset threshold is determined as the wake influence range of the wind turbine. The first preset threshold can be set based on engineering experience, for example, 0.05. Suppose within the 2D range downstream of the first wind turbine, the relative velocity deficit is greater than 0.05, then this region is determined as the wake influence range.

[0164] Within the wake influence range, a power curve model of the wind turbine is established. The power curve model describes the relationship between wind speed and wind turbine output power. Based on the power curve model and the calculated velocity deficit value, the wake loss power is calculated to obtain the power loss distribution of the wind turbine within the wake influence range. Suppose the output power of the first wind turbine is 2 MW when the incoming wind speed is 10 m / s, and the output power is 1.5 MW when the velocity deficit is 2 m / s, then the power loss is 0.5 MW.

[0165] Based on the wake influence range, the wind speed recovery rate is calculated along the center line of the wind wheel. The wind speed recovery rate is the ratio of the wind speed at any position on the center line to the incoming wind speed. For example, at 5D downstream of the first wind turbine, the wind speed recovery is 9 m / s, so the wind speed recovery rate is 9 / 10 = 0.9.

[0166] The downstream distance corresponding to the wind speed recovery rate greater than the second preset threshold is determined as the wake recovery distance. The second preset threshold can be set according to engineering experience, for example, 0.95. Assuming that the wind speed recovery rate reaches 0.95 at 8D downstream of the first wind turbine, the wake recovery distance is 8D.

[0167] The time sequence data of the wake influence range, the power loss distribution, and the wake recovery distance are stored in a wind turbine operation characteristic database for cooperative control of the wind turbines.

[0168] The beneficial effects of the method can be summarized in the following three aspects:

[0169] 1. Improved calculation accuracy: The adaptive grid refinement technique can automatically adjust the grid density according to the complexity of the wake field, ensuring calculation accuracy while improving calculation efficiency.

[0170] 2. Fine wake modeling: The Gaussian distribution function is used to describe the wake velocity deficit, which more accurately depicts the spatial distribution characteristics of the wake field, providing a reliable basis for subsequent power loss calculation and wake recovery distance determination.

[0171] 3. Optimization of wind farm control strategy: By storing the calculated wake characteristic data in the database, real-time data support can be provided for the cooperative control of wind turbines, thereby optimizing the overall output of the wind farm and improving power generation efficiency.

[0172] In an optional implementation, the cooperative control includes: when it is detected that the wake of an upstream wind turbine causes a power generation efficiency reduction of a downstream wind turbine to exceed a set threshold, controlling a yaw angle or a pitch angle of the upstream wind turbine to make the wake area of the upstream wind turbine avoid the downstream wind turbine, comprising:

[0173] Using a laser radar system to measure the three-dimensional wind speed distribution in the wake area of the upstream wind turbine in real time, and obtaining the wake characteristic data of the upstream wind turbine;

[0174] Collecting actual power output data of the downstream wind turbine through a wind farm monitoring system, comparing the actual power output data with theoretical power output data, and calculating the power generation efficiency loss of the downstream wind turbine;

[0175] When the power generation efficiency loss exceeds a preset loss threshold, based on the wake characteristic data and environmental parameters, using a dynamic programming algorithm to calculate an optimal yaw angle of the upstream wind turbine;

[0176] Adjusting the yaw of the upstream wind turbine according to the optimal yaw angle, and when the power generation efficiency loss after the yaw adjustment still exceeds the preset loss threshold, calculating an optimal pitch angle of the upstream wind turbine;

[0177] The yaw angle and the pitch angle of the upstream wind turbine are coordinated and adjusted by using a fuzzy P ID algorithm until the power generation efficiency loss of the downstream wind turbine is lower than the preset loss threshold.

[0178] The recovery degree of the power generation efficiency of the downstream wind turbine, the power generation loss of the upstream wind turbine, and the equipment load condition are monitored in real time, and the control parameters are adjusted online based on the monitoring results.

[0179] The laser radar system is deployed in the wind farm to measure the three-dimensional wind speed distribution in the wake area of the upstream wind turbine in real time. The laser radar system can be a scanning laser radar or a pulsed laser radar, and its measurement range can cover the area where the downstream wind turbine is located. The data collected by the laser radar include the three-dimensional wind speed value, wind direction, and turbulence intensity of each measurement point. For example, the wind speed measured by the laser radar at a distance of 500 meters, 700 meters, and 900 meters from the upstream wind turbine is 6 m / s, 5 m / s, and 4 m / s, respectively, the wind direction is 30°, 32°, and 35°, respectively, and the turbulence intensity is 0.12, 0.15, and 0.18, respectively. These data constitute the wake characteristic data of the upstream wind turbine.

[0180] The wind farm monitoring system collects the operating data of each wind turbine in real time, including actual power output, wind speed, wind direction, rotor speed, pitch angle, and yaw angle. For example, the actual power output of the downstream wind turbine is 800 kW, and the corresponding wind speed is 7 m / s, and the wind direction is 30°. At the same time, according to the power curve of the wind turbine, the theoretical power output of the downstream wind turbine under the current wind speed and wind direction can be calculated, for example, 950 kW.

[0181] By comparing the actual power output data of the downstream wind turbine with its theoretical power output data, the power generation efficiency loss of the downstream wind turbine is calculated. For example, the actual power output is 800 kW, and the theoretical power output is 950 kW, and the power generation efficiency loss is (950-800) / 950=15.8%.

[0182] The preset loss threshold is set according to the specific situation of the wind farm, for example, set to 10%. When the calculated power generation efficiency loss of the downstream wind turbine exceeds the preset 10% loss threshold (for example, 15.8%>10%), the wake control program is started.

[0183] Based on the wake characteristics data obtained by the laser radar (e.g. wind speed 6 m / s, 5 m / s, 4 m / s; wind direction 30°, 32°, 35°; turbulence intensity 0.12, 0.15, 0.18) and environmental parameters (e.g. ambient temperature, atmospheric pressure, air density), the optimal yaw angle of the upstream wind turbine is calculated using a dynamic programming algorithm. The goal of the dynamic programming algorithm is to find a yaw angle that minimizes the power generation efficiency loss of the downstream wind turbine, while also considering the power generation loss of the upstream wind turbine itself. For example, the calculated optimal yaw angle is 5°.

[0184] According to the calculated optimal yaw angle (e.g. 5°), the upstream wind turbine is adjusted in yaw.

[0185] After the yaw adjustment, the wind farm monitoring system will again collect the actual power output data of the downstream wind turbine and calculate its power generation efficiency loss. If the power generation efficiency loss still exceeds the pre-set loss threshold (e.g. 12%>10%), the optimal pitch angle of the upstream wind turbine needs to be further calculated. For example, the calculated optimal pitch angle is 2°.

[0186] The fuzzy PID algorithm is used to coordinate the adjustment of the yaw angle and pitch angle of the upstream wind turbine. The fuzzy PID algorithm adjusts the yaw angle and pitch angle of the upstream wind turbine in real time according to the power generation efficiency loss of the downstream wind turbine until the power generation efficiency loss is lower than the pre-set loss threshold (e.g. 8%<10%).

[0187] The wind farm monitoring system monitors the recovery degree of the power generation efficiency of the downstream wind turbine, the power generation loss of the upstream wind turbine, and the equipment load condition (e.g. blade bending moment, tower vibration) in real time. Based on the monitoring results, the parameters of the fuzzy PID controller are adjusted online, such as adjusting the proportional coefficient, integral coefficient and differential coefficient, to improve the control effect and ensure the safety of the equipment.

[0188] Advantages:

[0189] 1. Improve the overall power generation efficiency of the wind farm: By coordinating the control of the yaw angle and pitch angle of the upstream wind turbine, the impact of the wake on the downstream wind turbine can be effectively reduced, thereby improving the overall power generation efficiency of the entire wind farm.

[0190] 2. Reduce the operation and maintenance cost of the wind turbine: By optimizing the control strategy, the fatigue load of the wind turbine can be reduced, thereby prolonging its service life and reducing maintenance costs.

[0191] 3. Intelligent control: Advanced technologies such as laser radar, dynamic programming algorithm and fuzzy PID control are used to realize intelligent control of the wind turbine, improving the automation level of the wind farm.

[0192] In an optional embodiment, dynamically adjusting the rotational speed of each wind turbine in the wind turbine array according to the wake impact model to optimize the overall power generation efficiency of the wind turbine array includes:

[0193] Using a wind tower and a lidar system to collect three-dimensional wind field data of the wind farm, the three-dimensional wind field data is input into the Jensen wake model to establish a dynamic wake characteristic model that takes into account wind speed, turbulence intensity, and atmospheric stability;

[0194] Calculating a wake influence coefficient matrix between each wind turbine in a wind turbine array based on the dynamic wake characteristic model, wherein the wake influence coefficient matrix represents the wake interaction strength between any two wind turbines;

[0195] Inputting the wake influence coefficient matrix into the wind turbine array power output model to establish a mapping relationship between the speed of each wind turbine and the total power generation of the wind farm, wherein the mapping relationship simultaneously considers the impact of the speed on the power generation efficiency and wake characteristics of each wind turbine;

[0196] A multi-objective optimization problem is constructed based on the mapping relationship, and an improved particle swarm algorithm with an adaptive weight factor is used to solve the multi-objective optimization problem to obtain an optimal speed distribution scheme for each wind turbine;

[0197] Dividing the wind turbine array into a plurality of control sub-areas, executing the optimal speed distribution scheme in each of the control sub-areas, and collecting operating status data of each wind turbine in real time through a SCADA system;

[0198] The speed control effect is evaluated based on the operating status data. When a deviation between the actual power generation efficiency and the expected efficiency is detected, the parameters of the dynamic wake characteristic model are corrected online and the optimal speed distribution scheme is recalculated.

[0199] Three-dimensional wind field data collection and dynamic wake characteristic model development at a wind farm. Using multiple wind towers and a lidar system deployed within the wind farm, real-time data on wind speed, direction, turbulence intensity, and atmospheric stability is collected. The towers provide long-term, stable wind field data at a fixed location, while the lidar system can scan wind field information over a wider area. Combining the two yields more complete three-dimensional wind field data. For example, suppose a tower measures a wind speed of 8 m / s at a certain location, with a northerly wind direction, a turbulence intensity of 0.12, and an atmospheric stability level of 2. The lidar system then scans the area to obtain wind speed and direction distribution at different heights. This data is then fed into the Jensen wake model, which then considers the effects of wind speed, turbulence intensity, and atmospheric stability on wake attenuation and expansion to develop a dynamic wake characteristic model. This model dynamically predicts the propagation and impact range of wakes based on real-time wind field data.

[0200] Calculate the wake influence coefficient matrix based on the dynamic wake characteristics model. According to the dynamic wake characteristics model established in the previous step, calculate the wake influence coefficient matrix between each wind turbine in the wind turbine array. This matrix represents the strength of the wake interaction between any two wind turbines. For example, assuming there are 10 wind turbines in the wind farm, a 10x10 matrix needs to be calculated. Each element in the matrix represents the degree of wake influence of one wind turbine on another, with a value ranging from 0 to 1, with a larger value indicating stronger wake influence. For example, the value of the element in the 3rd row and 5th column of the matrix is 0.25, indicating that the wake influence coefficient of the 3rd wind turbine on the 5th wind turbine is 0.25.

[0201] Establish the mapping relationship between wind turbine speed and total power generation of the wind farm. Input the wake influence coefficient matrix calculated in the previous step into the wind turbine array power output model to establish the mapping relationship between the speed of each wind turbine and the total power generation of the wind farm. This mapping relationship takes into account the influence of speed on both single-machine power generation efficiency and wake characteristics. For example, assuming that the speed of a certain wind turbine is 10 rpm, its single-machine power generation is 2 MW, but due to the influence of the wake of the upstream wind turbine, the actual power generation is reduced to 1.8 MW. By adjusting the speed of this wind turbine, its own power generation efficiency and the influence on the downstream wind turbine can be changed, thereby affecting the total power generation of the entire wind farm.

[0202] Solve the multi-objective optimization problem using an improved particle swarm algorithm to obtain the optimal speed distribution scheme. Based on the mapping relationship established in the previous step, construct a multi-objective optimization problem, with the goal of maximizing the total power generation of the wind farm and minimizing the fatigue load of the wind turbine. Use an improved particle swarm algorithm with adaptive weight factors to solve this multi-objective optimization problem. Particle swarm optimization is an optimization algorithm that simulates the foraging behavior of a bird flock, searching for the optimal solution through iterative search. The adaptive weight factor can dynamically adjust the weight according to the convergence of the algorithm, improving the search efficiency and accuracy of the algorithm. Finally, the optimal speed distribution scheme for each wind turbine is obtained. For example, after optimization calculation, the optimal speeds of the 10 wind turbines are: 10 rpm, 11 rpm, 9 rpm, 12 rpm, 8 rpm, 10 rpm, 11 rpm, 9 rpm, 12 rpm, 8 rpm.

[0203] Divide the control sub-area and execute the optimal speed distribution scheme. Divide the wind turbine array into multiple control sub-areas, each containing several wind turbines. Execute the optimal speed distribution scheme obtained in the previous step in each control sub-area. For example, divide the 10 wind turbines into two control sub-areas, each containing 5 wind turbines. Collect the operating state data of each wind turbine in real time through the SCADA system, including speed, power, wind speed, and other information.

[0204] Based on the operating state data, the speed control effect is evaluated and online correction is performed. Based on the operating state data collected by the SCADA system, the speed control effect is evaluated. The actual power generation efficiency is compared with the expected efficiency, and when a deviation is detected between the two, the parameters of the dynamic wake characteristics model are corrected online. For example, if the actual power generation efficiency is lower than the expected efficiency, it may be necessary to adjust the decay coefficient or the expansion coefficient in the wake model. The optimal speed distribution scheme is recalculated, and the control instructions are updated to achieve closed-loop control of the wind turbine speed.

[0205] Advantages:

[0206] 1. Improve the power generation efficiency of the wind farm: by dynamically adjusting the speed of the wind turbines, the wake loss can be effectively reduced, and the overall power generation efficiency of the wind turbine array can be improved.

[0207] 2. Reduce the fatigue load of the wind turbine: the optimized speed distribution scheme can more evenly distribute the load of the wind turbine, thereby reducing the fatigue load of the wind turbine and prolonging its service life.

[0208] 3. Enhance the adaptability of the wind farm: this method can dynamically adjust the speed of the wind turbine according to the real-time wind conditions, enhance the adaptability of the wind farm to complex wind conditions, and improve the utilization rate of wind energy.

[0209] In an optional embodiment, a multi-objective optimization problem is constructed based on the mapping relationship, and an improved particle swarm algorithm with adaptive weight factors is used to solve the multi-objective optimization problem to obtain the optimal speed distribution scheme for each wind turbine, including:

[0210] A multi-objective optimization model of the wind turbine array is established, including a total power generation of the wind farm objective function, a wake interference objective function, and a unit load objective function. The total power generation of the wind farm objective function is the sum of the output power of each wind turbine, the wake interference objective function is the product of the wake influence coefficient and the speed loss value between each wind turbine, and the unit load objective function is the maximum load value of each wind turbine.

[0211] Based on the multi-objective optimization model, a particle coding scheme of the improved particle swarm algorithm is constructed, the speed of each wind turbine is taken as an optimization variable, the particle swarm is initialized, and the objective function value corresponding to each particle is calculated.

[0212] A nonlinear decreasing strategy is used to calculate the inertia weight of the improved particle swarm algorithm, the inertia weight is calculated by the power function of the ratio of the current iteration number to the maximum iteration number, and the exponent of the power function is a nonlinear adjustment factor.

[0213] The distance between each particle and the global optimal solution is calculated, and the individual learning factor and the social learning factor are dynamically calculated based on the distance by an exponential function; when the distance is large, the individual learning factor is increased to strengthen the local search ability, and when the distance is small, the social learning factor is increased to speed up the convergence speed;

[0214] The velocity and position of the particle are updated according to the inertia weight, the individual learning factor and the social learning factor, and the objective function value of the updated particle is calculated;

[0215] The non-dominated sorting of the updated particle swarm is performed to obtain the rank of each particle, and the crowding distance of the particles in the same rank is calculated; and the high-quality particles are selected based on the rank and the crowding distance to form a new generation of population;

[0216] The improvement rate of the objective function of the new generation of population is judged, and when the improvement rate of the objective function of the new generation of population is lower than the set threshold value or reaches the maximum iteration number for a continuous preset number of generations, the rotating speed configuration scheme corresponding to the global optimal particle is taken as the optimal rotating speed distribution scheme of each wind turbine generator unit; if the convergence condition is not met, the step of calculating the inertia weight is returned to continue iterative optimization.

[0217] The wind turbine array optimal rotating speed distribution method aims to improve the overall power generation efficiency of the wind farm. The method considers three objectives: total power generation of the wind farm, wake interference, and unit load, and uses an improved particle swarm optimization algorithm for solution.

[0218] Firstly, a multi-objective optimization model of the wind turbine array is established. The objectives of the model include maximizing the total power generation of the wind farm, minimizing the wake interference, and minimizing the unit load. The total power generation objective is defined as the sum of the output power of all wind turbine generators. The wake interference objective is defined as the sum of the product of the wake influence coefficient and the corresponding speed loss value between all wind turbine generators. The unit load objective is defined as the maximum load value among all wind turbine generators. For example, for a wind farm containing 10 wind turbine generators, the output power of each generator needs to be calculated and summed to obtain the total power generation; the wake influence coefficient and the speed loss value between each pair of generators are calculated, and the sum of all products is summed to obtain the wake interference; the maximum load value among the 10 generators is found as the unit load objective.

[0219] Next, based on the established multi-objective optimization model, the particle coding scheme of the improved particle swarm optimization algorithm is constructed. The rotating speed of each wind turbine generator is taken as the optimization variable, and the coding of the particle is constructed accordingly. For example, for 10 wind turbine generators, each particle is a vector containing 10 rotating speed values. Then, the particle swarm is initialized, a certain number of particles are randomly generated, and the three objective function values corresponding to each particle are calculated. For example, 50 particles are generated, each representing a rotating speed distribution scheme, and then the total power generation, wake interference, and unit load corresponding to each scheme are calculated.

[0220] In order to improve the convergence speed and global search ability of the algorithm, a nonlinear decreasing strategy is used to calculate the inertia weight of the improved particle swarm optimization algorithm. The inertia weight is calculated by the power function of the ratio of the current iteration number to the maximum iteration number, and the exponent of the power function is the nonlinear adjustment factor. For example, if the maximum iteration number is set to 100 and the nonlinear adjustment factor is set to 2, the inertia weight at the 50th iteration is (50 / 100)^2 = 0.25.

[0221] In each iteration process, the distance between each particle and the global optimal solution is calculated. Based on the distance, the individual learning factor and the social learning factor are dynamically calculated by the exponential function. When the distance is large, the individual learning factor is increased to enhance the local search ability; when the distance is small, the social learning factor is increased to speed up the convergence speed. For example, if the distance between a particle and the global optimal solution is 10, the individual learning factor may be 2.5 and the social learning factor may be 1.5; when the distance is 1, the individual learning factor may be 1.5 and the social learning factor may be 2.5.

[0222] According to the calculated inertia weight, individual learning factor and social learning factor, the speed and position of the particle are updated, and the objective function value of the updated particle is calculated. For example, if the current speed of a particle is 10, the speed is 1, the inertia weight is 0.8, the individual learning factor is 2, the social learning factor is 1.5, the individual optimal speed is 12, and the global optimal speed is 11, then the updated speed is 0.81 + 2(12-10) + 1.5(11-10) = 5.3, and the updated speed is 10 + 5.3 = 15.3.

[0223] The updated particle swarm is non-dominantly sorted to obtain the rank of each particle, and the crowding distance of the particles in the same rank is calculated. Based on the rank and the crowding distance, high-quality particles are selected to form a new generation of population. For example, particles with high rank are preferentially selected, and in the same rank, particles with large crowding distance are selected.

[0224] The improvement rate of the objective function of the new generation of population is judged. When the improvement rate of the objective function is lower than the set threshold value for a continuous preset number of generations or the maximum iteration number is reached, the speed configuration scheme corresponding to the global optimal particle is configured as the optimal speed distribution scheme of each wind turbine. For example, if the preset number of generations is set to 10 and the threshold value is set to 0.01, if the improvement rate of 10 generations is less than 0.01, or the iteration number reaches 100, the iteration is stopped. If the convergence condition is not met, the step of calculating the inertia weight is returned to continue the iteration optimization.

[0225] The beneficial effects of this method are reflected in three aspects:

[0226] 1. Improve wind farm power generation efficiency: By optimizing the rotational speed distribution scheme of wind turbines, the total power generation of the wind farm can be maximized considering the wake interference and turbine load, thereby improving the utilization efficiency of wind energy.

[0227] 2. Reduce turbine load: The method takes the turbine load as one of the optimization objectives, which can effectively reduce the maximum load value of the turbine, prolong the service life of the turbine, and reduce the maintenance cost.

[0228] 3. Intelligent control: The method uses an improved particle swarm algorithm to automatically search for the optimal rotational speed distribution scheme without human intervention, realizing intelligent control of wind turbines.

[0229] Figure 2 The structure diagram of the wind turbine array wake management system of the Doppler laser wind measurement embodiment of the present application is shown in Figure 2 The system comprises:

[0230] A first unit is used to deploy a Doppler laser radar in each wind turbine of a wind turbine array to collect real-time wind speed data, wind direction data and turbulence intensity data in front and back directions of each wind turbine in the wind turbine array; the Doppler laser radar calculates the real-time wind speed data, wind direction data and turbulence intensity data based on the Doppler frequency shift of the scattered light after the laser beam is reflected by air particles; the Doppler laser radar sets a wind measurement sampling point every 20 meters along the axis of the laser beam, and the wind measurement distance is 40-200 meters;

[0231] A second unit is used to establish a wake influence model of the wind turbine array according to the real-time wind speed data, wind direction data and turbulence intensity data; the wake influence model includes the wake influence coefficient, wake attenuation coefficient and wake diffusion coefficient of the upstream wind turbine on the downstream wind turbine; the wake influence range, wake loss power and wake recovery distance of each wind turbine in the wind turbine array are calculated through the wake influence model;

[0232] A third unit is used to cooperatively control the wind turbine array based on the wake influence range, wake loss power and wake recovery distance; the cooperative control includes: when it is detected that the wake of the upstream wind turbine causes the power generation efficiency of the downstream wind turbine to decrease by more than a set threshold, the yaw angle or pitch angle of the upstream wind turbine is controlled to make the wake area of the upstream wind turbine avoid the downstream wind turbine; at the same time, the rotational speed of each wind turbine in the wind turbine array is dynamically adjusted according to the wake influence model to optimize the overall power generation efficiency of the wind turbine array.

[0233] The third aspect of the embodiment of the present application is

[0234] An electronic device is provided, comprising:

[0235] a processor;

[0236] a memory for storing processor-executable instructions;

[0237] wherein the processor is configured to invoke the instructions stored by the memory to perform the method as described above.

[0238] A fourth aspect of the embodiments of the present application,

[0239] A computer readable storage medium is provided, which stores computer program instructions, and the computer program instructions are executed by a processor to implement the method as described above.

[0240] The present application can be a method, an apparatus, a system, and / or a computer program product. The computer program product can include a computer readable storage medium having computer readable program instructions stored therein, the computer readable program instructions being executable by a computer processor to perform aspects of the present application.

[0241] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

Claims

1. A method of wind turbine array wake management using Doppler lidar, characterized in that, The method comprises the following steps: Deploying a Doppler laser radar in each wind turbine of a wind turbine array to collect real-time wind speed data, wind direction data and turbulence intensity data in front and back directions of each wind turbine in the wind turbine array; The Doppler laser radar calculates the real-time wind speed data, wind direction data and turbulence intensity data based on the Doppler frequency shift of scattered light reflected by air particles after emitting a laser beam and receiving the scattered light; the Doppler laser radar sets a wind sampling point every 20 meters along the axial direction of the laser beam, and the wind measurement distance is 40-200 meters; According to the real-time wind speed data, wind direction data and turbulence intensity data, a wake influence model of the wind turbine array is established; the wake influence model includes a wake influence coefficient, a wake attenuation coefficient and a wake diffusion coefficient of an upstream wind turbine on a downstream wind turbine; The wake influence range, wake loss power and wake recovery distance of each wind turbine in the wind turbine array are calculated by the wake influence model, including: A cylindrical coordinate system with the center of the wind wheel as the origin and the incoming flow direction as the axial direction is established, an initial calculation grid is constructed in the cylindrical coordinate system, a grid encryption coefficient is determined by calculating the velocity gradient, and the calculation grid is adaptively encrypted based on the grid encryption coefficient to obtain an optimized calculation grid; At each grid point of the optimized calculation grid, a velocity deficit value of a downstream position point of the wind turbine is calculated using a velocity deficit Gaussian distribution function, and the ratio of the velocity deficit value to the incoming flow speed is determined as the relative velocity deficit; Based on the relative velocity deficit, by scanning the grid points in the calculation grid, a spatial region with a relative velocity deficit greater than a first preset threshold value is determined as the wake influence range of the wind turbine; Within the wake influence range, a power curve model of the wind turbine is established, and the wake loss power is calculated based on the power curve model and the velocity deficit value to obtain the power loss distribution of the wind turbine within the wake influence range; Based on the wake influence range, the wind speed recovery rate is calculated along the center line of the wind wheel, and the downstream distance corresponding to the wind speed recovery rate greater than a second preset threshold value is determined as the wake recovery distance, wherein the wind speed recovery rate is the ratio of the wind speed at any position on the center line to the incoming flow speed; The time sequence data of the wake influence range, the power loss distribution and the wake recovery distance are stored in a wind turbine operation characteristic database for collaborative control of the wind turbine; Based on the wake influence range, wake loss power and wake recovery distance, the wind turbine array is collaboratively controlled; the collaborative control includes: when it is detected that the wake of an upstream wind turbine causes the power generation efficiency of a downstream wind turbine to decrease by more than a set threshold value, the yaw angle or pitch angle of the upstream wind turbine is controlled to make the wake area of the upstream wind turbine avoid the downstream wind turbine; at the same time, the rotational speed of each wind turbine in the wind turbine array is dynamically adjusted according to the wake influence model to achieve optimal overall power generation efficiency of the wind turbine array.

2. The method of claim 1, wherein, The Doppler laser radar calculates the real-time wind speed data, wind direction data and turbulence intensity data based on Doppler frequency shift of scattered light of the laser beam reflected by air particles, including: The Doppler laser radar emits a narrow-band laser beam with a wavelength of 1.5 μm, the reference light frequency of the Doppler laser radar is offset by 80 MHz relative to the frequency of the narrow-band laser beam, and scattered light signals of the narrow-band laser beam scattered by air particles are collected; the scattered light signals and the reference light are heterodyne mixed on a photodetector to obtain beat signals; A velocity azimuth angle scanning scheme is adopted, the laser beam of the Doppler laser radar is controlled to rotate and scan in the horizontal plane at an elevation angle of 30 degrees, the scattered light signals are collected every 30 degrees of azimuth angle, and the beat signals of 12 azimuth angles are obtained within a range of 360 degrees; the beat signals are subjected to fast Fourier transform to obtain Doppler frequency spectrum, and radial wind speed is calculated based on the peak position of the Doppler frequency spectrum; Wavelet transform is used for adaptive threshold noise reduction of the Doppler frequency spectrum, and reliable data is selected in real time by monitoring signal quality indicators; more than 100 valid samples are accumulated for each measurement point, the horizontal plane wind speed vector is reconstructed by least square fitting, and wind speed and wind direction angle are obtained; the ratio of the standard deviation to the average value of the radial wind speed is calculated to obtain the turbulence intensity; the time resolution of the wind field parameter solving method is 1 second, and the spatial resolution is 20 meters.

3. The method of claim 1, wherein, According to the real-time wind speed data, wind direction data and turbulence intensity data, a wake influence model of the wind turbine array is established; the wake influence model includes a wake influence coefficient, a wake attenuation coefficient and a wake diffusion coefficient of an upstream wind turbine on a downstream wind turbine, including: Real-time wind speed data in the downstream area of the wind turbine is collected, and a velocity deficit Gaussian distribution function is established based on the real-time wind speed data: ， where V(x, r) represents the wind speed at a downstream distance x and a radial distance r, V∞ is the incoming flow wind speed, CT is the thrust coefficient, α(x) is the wake influence coefficient, and σ(x) is the wake diffusion width; Wind farm turbulence intensity data is obtained, and a wake attenuation coefficient is calculated based on the turbulence intensity data: ， where β is the wake attenuation coefficient, and TI is the turbulence intensity; The wake attenuation coefficient β is substituted into the wake influence coefficient calculation formula: ， to obtain the wake influence coefficient α(x); where x is the downstream distance, and D is the rotor diameter; A wake diffusion model considering wind direction deviation is established based on the wind direction data: ， where k is the wake diffusion coefficient, and the wake diffusion model is substituted into the velocity deficit Gaussian distribution function; For the case that a downstream wind turbine is simultaneously affected by the wakes of multiple upstream wind turbines, a wake superposition model is established based on the principle of conservation of momentum: where δp is a parameter correction amount, and γ is a learning rate; ， where ΔV total is the velocity deficit after superposition, ΔV i is the velocity deficit caused by a single wake; Real-time acquisition of lidar wind speed measurements V measured Comparison of the lidar wind speed measurements with model predicted wind speed values V model Calculation of parameter correction quantities: ， The wake influence coefficient, the wake attenuation coefficient and the wake diffusion coefficient are adaptively corrected based on the parameter correction amount. ​ 4. The method of claim 1, wherein, The cooperative control comprises: when it is detected that the wake of an upstream wind turbine causes the power generation efficiency of a downstream wind turbine to decrease by more than a set threshold, controlling the yaw angle or the pitch angle of the upstream wind turbine to avoid the wake area of the upstream wind turbine from the downstream wind turbine comprises: Real-time measurement of three-dimensional wind speed distribution in the wake area of the upstream wind turbine by a laser radar system to obtain the wake characteristic data of the upstream wind turbine; Collection of actual power output data of the downstream wind turbine by a wind farm monitoring system, comparison of the actual power output data with theoretical power output data, and calculation of the power generation efficiency loss of the downstream wind turbine; When the power generation efficiency loss exceeds a preset loss threshold, based on the wake characteristic data and environmental parameters, a dynamic programming algorithm is used to calculate the optimal yaw angle of the upstream wind turbine; According to the optimal yaw angle, the upstream wind turbine is adjusted in yaw, and when the power generation efficiency loss after the yaw adjustment still exceeds the preset loss threshold, the optimal pitch angle of the upstream wind turbine is calculated; The yaw angle and the pitch angle of the upstream wind turbine are adjusted by a fuzzy PID algorithm until the power generation efficiency loss of the downstream wind turbine is lower than the preset loss threshold; Real-time monitoring of the recovery degree of the power generation efficiency of the downstream wind turbine, the power generation loss of the upstream wind turbine, and the equipment load condition, and online optimization and adjustment of the control parameters based on the monitoring results.

5. The method of claim 1, wherein, Meanwhile, according to the dynamic adjustment of the wake influence model, the rotational speed of each wind turbine in the wind turbine array is adjusted to achieve the optimization of the overall power generation efficiency of the wind turbine array comprises: Collection of three-dimensional wind field data of the wind farm by a wind measurement tower and a laser radar system, input of the three-dimensional wind field data into a Jensen wake model, and establishment of a dynamic wake characteristic model considering wind speed, turbulence intensity, and atmospheric stability; Calculation of a wake influence coefficient matrix between each wind turbine in the wind turbine array based on the dynamic wake characteristic model, the wake influence coefficient matrix representing the wake interaction strength between any two wind turbines; Input of the wake influence coefficient matrix into a wind turbine array power output model to establish a mapping relationship between the rotational speed of each wind turbine and the total power generation of the wind farm, the mapping relationship considering the influence of the rotational speed on the power generation efficiency of a single machine and the wake characteristics; Based on the mapping relationship, a multi-objective optimization problem is constructed, and an improved particle swarm algorithm with an adaptive weight factor is used to solve the multi-objective optimization problem to obtain an optimal rotational speed distribution scheme for each wind turbine; The wind turbine array is divided into multiple control sub-regions, and the optimal rotational speed distribution scheme is executed in each control sub-region, and the operating state data of each wind turbine is collected in real time by a SCADA system; Based on the operating state data, the effect of rotational speed control is evaluated, and when a deviation is detected between the actual power generation efficiency and the expected efficiency, the parameters of the dynamic wake characteristic model are corrected online, and the optimal rotational speed distribution scheme is recalculated.

6. The method of claim 5, wherein, Based on the mapping relationship, a multi-objective optimization problem is constructed, and an improved particle swarm algorithm with an adaptive weight factor is used to solve the multi-objective optimization problem to obtain an optimal speed distribution scheme of each wind turbine, including: A multi-objective optimization model of the wind turbine array is established, including a total power generation target function of the wind farm, a wake interference target function, and a unit load target function; wherein the total power generation target function of the wind farm is the sum of the output power of each wind turbine, the wake interference target function is the product of the wake influence coefficient and the speed loss value between each wind turbine, and the unit load target function is the maximum load value of each wind turbine; Based on the multi-objective optimization model, a particle coding scheme of the improved particle swarm algorithm is constructed, the speed of each wind turbine is taken as an optimization variable, the particle swarm is initialized, and the target function value corresponding to each particle is calculated; The inertia weight of the improved particle swarm algorithm is calculated using a nonlinear decreasing strategy, the inertia weight is calculated by a power function of the ratio of the current iteration number to the maximum iteration number, and the exponent of the power function is a nonlinear adjustment factor; The distance between each particle and the global optimal solution is calculated, and the individual learning factor and the social learning factor are dynamically calculated based on the distance; when the distance is large, the individual learning factor is increased to enhance the local search ability, and when the distance is small, the social learning factor is increased to speed up the convergence speed; The speed and position of the particle are updated according to the inertia weight, the individual learning factor and the social learning factor, and the target function value of the updated particle is calculated; The updated particle swarm is non-dominantly sorted to obtain the rank of each particle, and the crowding distance of the particle is calculated in the same rank; based on the rank and the crowding distance, a high-quality particle is selected to form a new generation population; The target function improvement rate of the new generation population is judged, when the target function improvement rate of a continuous preset number of generations is lower than a set threshold or reaches a maximum iteration number, the speed configuration scheme corresponding to the global optimal particle is taken as the optimal speed distribution scheme of each wind turbine; if the convergence condition is not met, the inertia weight calculation step is returned to continue iterative optimization.

7. A wind turbine array wake management system using Doppler lidar for implementing the method of any of the preceding claims 1-6, characterized in that, Including: A first unit is configured to deploy a Doppler laser radar in each wind turbine of a wind turbine array to collect real-time wind speed data, wind direction data and turbulence intensity data in front and back directions of each wind turbine in the wind turbine array; The Doppler laser radar calculates the real-time wind speed data, wind direction data and turbulence intensity data based on the Doppler frequency shift of scattered light reflected by air particles after emitting a laser beam and receiving the scattered light; the Doppler laser radar sets a wind measurement sampling point every 20 meters along the axial direction of the laser beam, and the wind measurement distance is 40-200 meters. a second unit configured to establish a wake effect model of the wind turbine array according to the real-time wind speed data, the wind direction data, and the turbulence intensity data; the wake effect model including a wake effect coefficient of an upstream wind turbine on a downstream wind turbine, a wake attenuation coefficient, and a wake diffusion coefficient; and the wake effect model being used to calculate a wake effect range, a wake loss power, and a wake recovery distance of each wind turbine in the wind turbine array; a third unit configured to perform a coordinated control on the wind turbine array based on the wake effect range, the wake loss power, and the wake recovery distance; the coordinated control including: when detecting that a wake of an upstream wind turbine causes a power generation efficiency of a downstream wind turbine to decrease by more than a set threshold, controlling a yaw angle or a pitch angle of the upstream wind turbine so that a wake area of the upstream wind turbine avoids the downstream wind turbine; and dynamically adjusting a rotating speed of each wind turbine in the wind turbine array according to the wake effect model to achieve an overall power generation efficiency optimization of the wind turbine array.

8. An electronic device, comprising: comprise: a processor; a memory for storing processor-executable instructions; wherein the processor is configured to invoke the instructions stored in the memory to perform the method of any one of claims 1 to 6.

9. A computer-readable storage medium having stored thereon computer program instructions, wherein, the computer program instructions, when executed by the processor, implement the method of any one of claims 1 to 6.

Images