Wind power plant energy management method based on laser radar wind measurement

Abstract

The invention relates to the technical field of wind power generation, and discloses a wind power plant energy management method based on laser radar wind measurement, comprising the following steps: S1, deploying laser radar equipment, and collecting real-time data of wind speed, wind direction and turbulence intensity in a wind power plant; s2, constructing a three-dimensional dynamic model of a wind field, and modeling wind speed, wind direction and turbulence distribution in combination with laser radar data and historical meteorological data; s3, performing short-term prediction on the future wind speed by using the deep learning model, and outputting minute-level to hour-level wind energy prediction data; s4, optimizing a dynamic scheduling strategy of the wind turbine generator based on the prediction data; and S5, docking an energy storage system, optimizing charging and discharging behaviors of energy storage equipment, and realizing smooth power output. According to the invention, by fusing the laser radar wind measurement data, the deep learning prediction model and the intelligent optimization algorithm, high-precision wind condition monitoring, dynamic scheduling optimization and power output stability improvement of the wind power plant are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of wind power generation, and in particular to a wind farm energy management method based on laser radar wind measurement. Background Art

[0002] As a clean and renewable energy source, wind power generation plays an important role in optimizing the global energy structure. Wind farm energy management methods are the core link in wind farm operation. Through the collection, prediction and processing of wind speed, wind direction and other wind information, combined with the optimization control of wind turbine scheduling and energy storage systems, efficient utilization of wind energy resources can be achieved.

[0003] Existing wind farm energy management methods usually rely on traditional meteorological towers, wind speed sensors or simplified physical models to monitor and predict wind conditions, and combine them with static scheduling strategies based on power curves to control wind turbine groups. This type of method has the advantages of low equipment cost and simple operation, and shows good wind energy utilization efficiency under flat terrain and single wind farm conditions. At the same time, the energy storage scheduling strategy based on statistical models can achieve a certain degree of power smoothing and reduce the impact of wind farms on the power grid.

[0004] However, existing technologies have shortcomings in complex terrain wind farms, regional wind farm coordination, wind turbine operation safety and dynamic adaptability; wind condition monitoring accuracy is insufficient, making it difficult to capture the turbulence and wake effects of complex wind farms; the real-time and accuracy of wind energy forecast results are low, affecting the efficiency of dynamic scheduling; abnormal wind condition identification and fault diagnosis functions are imperfect, increasing equipment operation risks; and insufficient coordination between regional wind farms limits the overall wind energy utilization efficiency. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a wind farm energy management method based on lidar wind measurement, which solves the problems in the prior art of insufficient wind condition monitoring accuracy under complex terrain and coordinated operation of multiple wind farms, low real-time and accuracy of short-term wind energy forecasting, low efficiency of wind turbine dynamic scheduling, and imperfect abnormal wind condition identification and equipment fault diagnosis functions.

[0006] To achieve the above objectives, the present invention is implemented through the following technical solutions: A wind farm energy management method based on laser radar wind measurement includes the following steps:

[0007] S1. Deploy lidar equipment to collect real-time data on wind speed, wind direction and turbulence intensity in the wind farm;

[0008] S2, build a three-dimensional dynamic model of the wind field, combine lidar data and historical meteorological data to model wind speed, wind direction and turbulence distribution;

[0009] S3. Use deep learning models to make short-term predictions of future wind speeds and output wind energy forecast data from minute to hour levels.

[0010] S4. Optimize the dynamic dispatch strategy of wind turbines based on forecast data;

[0011] S5. Connect to the energy storage system, optimize the charging and discharging behavior of the energy storage equipment, and achieve smooth power output;

[0012] S6. Integrate the operating data of multiple wind farms and optimize regional power synergy through virtual power plants;

[0013] S7. Detect abnormal wind conditions and diagnose the operating status of wind turbines, and trigger corresponding operating protection or alarm.

[0014] Preferably, the S1 comprises:

[0015] Deploy LiDAR equipment in key areas of the wind farm to ensure that the equipment measurement range covers the main operating areas of the wind farm;

[0016] Set the LiDAR measurement parameters, including measurement frequency, scanning angle, and measurement range;

[0017] Collect wind speed, wind direction and turbulence intensity data in real time and transmit the data to the data processing center.

[0018] Preferably, S2 includes:

[0019] Fusion of wind speed, wind direction and turbulence intensity data collected in real time by lidar with historical meteorological data;

[0020] Interpolation algorithms are used to interpolate multi-point data in the wind farm to construct a three-dimensional dynamic model of the wind farm;

[0021] The turbulence intensity distribution and wake effect in the wind field model are corrected to generate the optimized results of the wind speed distribution in the wind field.

[0022] Preferably, S3 includes:

[0023] Extract the time series features of real-time lidar data, including wind speed change trends, wind direction change trends, and turbulence change trends;

[0024] Short-term wind energy forecasting using deep learning models;

[0025] Output prediction results for dynamic scheduling and energy storage optimization.

[0026] Preferably, S4 includes:

[0027] Dynamically adjust wind turbine operating parameters, including pitch angle and speed ratio;

[0028] Calculate the optimal value of wind turbine power output using optimization algorithm;

[0029] Select appropriate wind turbine start and stop strategies under different wind conditions to maximize wind energy utilization.

[0030] Preferably, S5 includes:

[0031] Predict the difference between wind power output and grid power demand;

[0032] Adjust the charging and discharging strategy of the energy storage system to make the power output smooth;

[0033] Dynamically optimize energy storage power distribution according to the status of the energy storage system and load demand.

[0034] Preferably, S6 includes:

[0035] Integrate real-time operating data from multiple wind farms;

[0036] Construct a regional wind energy collaborative prediction model to predict the wind power output capacity in the region;

[0037] The power output of regional wind farms is optimized and dispatched through virtual power plants.

[0038] Preferably, the S7 includes:

[0039] Real-time detection of abnormal wind conditions based on lidar wind speed sequence data and machine learning algorithms;

[0040] Triggering wind turbine protection mechanisms when turbulence, wind shear or extreme wind conditions are detected;

[0041] Combined with the fan operating data and power deviation, the fan operating status is diagnosed and a fault alarm is generated.

[0042] Preferably, the interpolation algorithm is a Kriging interpolation algorithm, and during the interpolation process, the interpolation weights are dynamically adjusted according to the terrain characteristics of the wind farm, so that the interpolation result is more suitable for the wind speed distribution under complex terrain.

[0043] Preferably, the short-term wind energy forecast adopts a deep learning algorithm based on time series, and the historical operation data and terrain characteristics of the wind farm are introduced into the model training process to improve the accuracy and real-time performance of the forecast results.

[0044] The present invention provides a wind farm energy management method based on laser radar wind measurement. It has the following beneficial effects:

[0045] 1. The present invention constructs a dynamic wind field model and an accurate short-term wind energy prediction model by real-time acquisition and integration of lidar data, historical meteorological data and terrain features, realizes efficient monitoring and prediction of wind speed, wind direction and turbulence distribution under complex wind conditions, and provides reliable data support for wind farm operation optimization.

[0046] 2. The present invention adopts intelligent optimization algorithms and deep learning technology to optimize wind turbine parameters, energy storage scheduling and regional wind farm coordination, thereby improving wind energy utilization, power output stability and regional resource integration efficiency, and supporting efficient operation of wind farms and grid adaptation.

[0047] 3. The present invention realizes active safety protection and equipment failure prediction of wind farms through real-time detection of abnormal wind conditions, wind turbine protection mechanism and status diagnosis function, reduces operating risks and maintenance costs, extends equipment service life, and ensures the safety and economic benefits of wind farm operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 The present invention is a flow chart of the method. DETAILED DESCRIPTION

[0049] The following will be combined with the drawings in the specification of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0050] Please see attached Figure 1 The embodiment of the present invention provides a wind farm energy management method based on laser radar wind measurement, comprising the following steps:

[0051] S1. Deploy lidar equipment to collect real-time data on wind speed, wind direction and turbulence intensity in the wind farm. Through the high-precision wind measurement of lidar, comprehensive monitoring of wind conditions in the wind farm can be achieved, providing accurate data support for subsequent wind farm modeling and energy management, and improving the response speed and reliability of the system;

[0052] S2. Construct a three-dimensional dynamic model of the wind farm. Combine the lidar data and historical meteorological data to model the wind speed, wind direction and turbulence distribution. The constructed three-dimensional dynamic model provides an accurate description of the wind conditions, laying the foundation for optimizing wind turbine operation, power prediction and scheduling decisions, and can improve the overall operating efficiency of the wind farm.

[0053] S3. Use deep learning models to make short-term predictions of future wind speeds and output minute- to hour-level wind energy forecast data. Accurate short-term wind speed forecasts reduce the uncertainty of wind farm operation, provide a reliable basis for wind turbine scheduling and energy storage optimization, and improve the stability of wind power output.

[0054] S4. Optimize the dynamic dispatch strategy of wind turbines based on the forecast data. The dynamic dispatch strategy can maximize the utilization of wind energy while reducing the impact of the wake effect on wind turbines, significantly improving the overall output power of the wind farm;

[0055] S5. Connect the energy storage system, optimize the charging and discharging behavior of the energy storage equipment, and achieve smooth power output. The optimized energy storage control achieves smooth power output, reduces the pressure of grid frequency regulation, and improves the economic benefits of the wind farm;

[0056] S6. Integrate the operating data of multiple wind farms and optimize regional power through virtual power plants. Virtual power plant optimization improves the overall operating efficiency of regional wind farms and enhances the ability of wind farms to participate in the power market;

[0057] S7. Detect abnormal wind conditions and diagnose the operating status of wind turbines, trigger corresponding operating protection or alarms, and improve the safety of wind farm operation through abnormal detection and diagnosis, reducing the risk of failure and unplanned downtime.

[0058] Please see attached Figure 1 In a preferred embodiment of the present invention, S1 includes:

[0059] LiDAR equipment is deployed in key areas of the wind farm to ensure that the equipment measurement range covers the main operating area of ​​the wind farm. LiDAR is deployed in key areas of the wind farm, preferably on the windward side, to improve the monitoring effect, thereby achieving comprehensive coverage of wind speed, wind direction and turbulence intensity in the wind farm, ensuring the spatial continuity and representativeness of the data, providing high-quality input for subsequent data processing and optimization, reducing monitoring blind spots in the wind turbine operating area, and improving the operation control accuracy of the wind farm;

[0060] Set the laser radar measurement parameters, including measurement frequency, scanning angle and measurement range. Set a higher frequency (10-20Hz) in areas with rapid wind speed changes (such as high turbulence areas); set a lower frequency (1-5Hz) in areas with stable wind conditions to balance measurement accuracy and data volume. Determine the horizontal scanning angle of the laser radar, covering the horizontal plane from 0° to 360°, and monitor the wind conditions of the wind turbine in all directions; set the vertical scanning range (such as -10° to +30°) to capture the wind speed distribution at different altitudes. The measurement parameters optimize the measurement capabilities of the laser radar, which can not only meet the real-time monitoring needs, but also avoid unnecessary data redundancy and improve the system operation efficiency.

[0061] Wind speed, wind direction and turbulence intensity data are collected in real time and transmitted to the data processing center. The lidar captures the dynamic changes of wind conditions through high-frequency sampling. After the real-time data is transmitted to the control center, it can support subsequent wind farm modeling, prediction and scheduling, and ensure the continuity and reliability of wind farm operation.

[0062] Please see attached Figure 1 In a preferred embodiment of the present invention, S2 includes:

[0063] The wind speed, wind direction and turbulence intensity data collected by the lidar in real time are integrated with the historical meteorological data. Data fusion is based on Bayesian theory to achieve the integration of multi-source data. The confidence of the lidar data and historical data is calculated, and the fusion weight is dynamically adjusted:

[0064] Wind speed Fusion of:

[0065]

[0066] Among them, w1 and w2 are confidence weights. Data fusion effectively improves the integrity and reliability of wind farm data, provides a more accurate input basis for the three-dimensional dynamic model of the wind farm, and helps to improve the accuracy of subsequent modeling and prediction;

[0067] The interpolation algorithm is used to interpolate the multi-point data in the wind farm and construct a three-dimensional dynamic model of the wind farm. First, the distribution of interpolation points in the wind farm is determined according to the measurement range of the lidar and the geographical layout of the wind farm. In addition, high-density interpolation points are set in key areas (such as turbulence-prone areas) in combination with the location of wind turbines and terrain characteristics. The Kriging interpolation algorithm is used to perform spatial interpolation of wind speed, wind direction and turbulence intensity. The formula is as follows:

[0068]

[0069] Where, v(x, y, z): wind speed value at the interpolation point; μ: global average wind speed; λ i : interpolation weight, determined by the covariance matrix; v i : Given the wind speed value at the measuring point, a three-dimensional dynamic model of the wind field is constructed based on the interpolation results to describe the distribution of wind speed, wind direction and turbulence intensity in space and time. The three-dimensional dynamic wind field model can accurately describe the wind conditions at different heights and locations in the wind farm, providing reliable input for wind turbine scheduling optimization and power prediction;

[0070] Correct the turbulence intensity distribution and wake effect in the wind field model to generate the optimized result of wind speed distribution in the wind field, where the turbulence intensity is corrected:

[0071] Calculating the turbulence intensity I in the interpolation model t :

[0072]

[0073] σ v : standard deviation of wind speed; Average wind speed;

[0074] Correct the abnormal turbulence intensity areas (such as turbulence intensity is too high or too low) to ensure the physical rationality of the model;

[0075] Wake Effect Correction:

[0076] The Jensen wake model is used to calculate the wind speed correction value in the wake area:

[0077]

[0078] Among them, v w (x): wind speed in the wake area; v h : free wind speed; a: induction factor; k: wake expansion coefficient; The ratio of the wake position to the wind turbine diameter is used to integrate the corrected turbulence intensity and wake effect results into the three-dimensional wind field model to generate an optimized wind speed distribution. The corrected model more realistically reflects the wind speed distribution and flow field characteristics in the wind farm, providing a reliable basis for the optimization of wind turbine operating parameters and scheduling strategies.

[0079] Please see attached Figure 1 In a preferred embodiment of the present invention, S3 includes:

[0080] Extract the time series features of real-time lidar data, including wind speed change trends, wind direction change trends, and turbulence change trends. The time series features of wind speed, wind direction, and turbulence contain information about the dynamic changes in the wind field. By extracting trends and statistical characteristics, the key patterns of wind condition changes can be captured, providing high-quality input for the prediction model. The extraction of time series features enhances the structure and usability of the data, allowing the wind energy prediction model to more accurately learn the time series patterns of wind condition changes, thereby improving the prediction accuracy;

[0081] Use deep learning models to forecast short-term wind energy. Select LSTM (Long Short-Term Memory Network) or Transformer model, input the time series features extracted in real time into the model, and output wind speed forecast results from minute to hour. The deep learning model can accurately capture the trend and fluctuation characteristics of wind conditions by learning the nonlinearity and temporal dependency in the time series, thereby achieving high-precision short-term wind energy forecasting, and thus providing accurate wind speed forecast results, providing a reliable basis for dynamic scheduling of wind farms and energy storage optimization.

[0082] Output prediction results for dynamic scheduling and energy storage optimization, convert wind speed prediction results into wind turbine power and energy storage requirements, and provide quantitative basis for dynamic scheduling and energy storage control. Real-time updated prediction results can adapt to rapid changes in wind conditions and ensure the continuity of wind farm operation. High-precision prediction results directly drive the dynamic optimization strategy of wind farms, improve wind energy utilization efficiency, reduce power output volatility, and improve wind power's adaptability to the grid.

[0083] Please see attached Figure 1 In a preferred embodiment of the present invention, S4 includes:

[0084] Dynamically adjust the operating parameters of the wind turbine, including the pitch angle and speed ratio. According to the real-time wind speed prediction value v(t), dynamically adjust the pitch angle β of the wind turbine to keep the wind turbine running at the optimal power point. When the wind speed is too high or too low, adjust the pitch angle to avoid overspeed operation or energy loss.

[0085] Speed ​​ratio optimization:

[0086] Calculate the best speed ratio in real time Where ω is the fan speed, R is the fan radius, and v is the wind speed;

[0087] Adjust the fan speed controller to maintain the best speed ratio and maximize the power coefficient C p ;

[0088] The pitch angle and speed ratio directly affect the power output and operating efficiency of the wind turbine. Through real-time adjustment, the wind turbine can always operate in the high-efficiency range, while avoiding equipment damage or efficiency reduction caused by extreme wind conditions, thereby optimizing the operating status of the wind turbine and extending the service life of the wind turbine;

[0089] The optimal value of the fan power output is calculated using the optimization algorithm, using the fan power curve formula:

[0090]

[0091] Where, P: air density; A: fan swept area; v: wind speed; C p : power coefficient;

[0092] With the goal of maximizing the wind turbine output power P, the optimization problem is constructed:

[0093] maxC p (λ, β)

[0094] The constraints are the physical limits of the speed ratio λ and the pitch angle β. The optimization algorithm can be used to quickly calculate the optimal operating parameters of the wind turbine, improve the accuracy of power output, and reduce unnecessary energy losses during wind turbine operation.

[0095] Select appropriate wind turbine start-stop strategies under different wind conditions to maximize wind energy utilization. The operating status of wind turbines under different wind conditions directly determines the efficiency of wind energy utilization and the safety of equipment. By formulating targeted start-stop strategies, equipment damage can be avoided while optimizing power generation. Appropriate wind turbine start-stop strategies effectively improve wind energy utilization efficiency, reduce mechanical losses of wind turbines, and ensure the safety of wind farms under extreme wind conditions.

[0096] Please see attached Figure 1 In a preferred embodiment of the present invention, S5 includes:

[0097] Predict the difference between wind power output and grid power demand, using the wind speed forecast value v(t+1), based on the wind turbine power curve formula Calculate the wind power output P at the future time t+1 wind (t+1), the power of each wind turbine is summarized to generate the predicted value of the total output power of the wind farm;

[0098] Collect the real-time load data L(t) of the power grid, and combine it with the power demand forecasting model (such as time series model or load forecasting model based on machine learning) to predict the power demand P of the power grid in the future wind (t+1);

[0099] Calculate the difference between wind power output and grid demand:

[0100] ΔP(t+1)=P wind (t+1)-P load (t+1)

[0101] The difference ΔP(t+1) is the input for subsequent energy storage system adjustments. By calculating the difference between wind power and grid demand, future power fluctuations and imbalances can be identified, and energy storage strategies can be planned in advance for smoothing. The difference calculation provides accurate input data for the energy storage system, which can significantly improve the pertinence and real-time performance of power smoothing and reduce the frequency regulation pressure on the grid caused by power fluctuations.

[0102] Adjust the charging and discharging strategy of the energy storage system to make the power output smooth. The energy storage system compensates for the fluctuation of wind power in real time through charging and discharging behavior, making the output power closer to the grid demand and avoiding grid fluctuations caused by insufficient or excessive power. The dynamic charging and discharging strategy of the energy storage system reduces the volatility of power output, improves the adaptability of wind farms to grid loads, and improves the quality and reliability of wind power grid connection.

[0103] According to the state of the energy storage system and the load demand, the energy storage power distribution is dynamically optimized. First, a multi-objective optimization model is constructed to balance the charging and discharging requirements of the energy storage system, the life of the energy storage equipment, and the priority of power fluctuation compensation. Then, the objective function is optimized:

[0104] minJ=α·SOC deviation +β·P fluctuation +γ·C lifetime

[0105] Among them, SOC deviation : state deviation of energy storage system; P fluctuation : Power fluctuation; C lifetime : The life cost of energy storage equipment, at the same time, give priority to meeting the load demand of the power grid, allocate more energy storage power to the period of large power fluctuations, and dynamically adjust the discharge rate priority when the energy storage capacity is insufficient to ensure the stability of the core load of the power grid; finally, use the real-time feedback of the energy storage system (such as current power, temperature and health status) to dynamically adjust the weight parameters α, β, γ of the optimization model;

[0106] The dynamic optimization allocation strategy improves the responsiveness and operating efficiency of the energy storage system, while extending the service life of the energy storage equipment and reducing operation and maintenance costs.

[0107] Please see attached Figure 1 In a preferred embodiment of the present invention, S6 includes:

[0108] Integrate the real-time operation data of multiple wind farms. By integrating the real-time operation data of multiple wind farms, complete wind power output and energy storage status information can be obtained within the region, thereby enhancing the coordination ability of regional wind farms and laying the foundation for unified management and optimized scheduling among multiple wind farms.

[0109] Construct a regional wind energy collaborative prediction model to predict the wind power output capacity in the region. The regional wind energy collaborative prediction model can accurately predict the wind power output capacity within the region by learning the spatial correlation and temporal change pattern between wind farms, thereby improving the accuracy and timeliness of regional wind energy prediction and providing a reliable basis for the collaborative optimization of regional wind farms.

[0110] Optimize the power output of regional wind farms through virtual power plants;

[0111] Specifically: The implementation method of optimizing the power output of regional wind farms through virtual power plants includes modeling multiple wind farms and energy storage systems as a virtual power plant (VPP) that participates in power dispatch as a single power generation unit;

[0112] Model input: power forecast value P of regional wind farm region (t+1), energy storage system status E SoC and grid load demand P load (t+1);

[0113] With the goal of maximizing economic benefits, construct an optimization function:

[0114]

[0115] Among them, P VPP (t): power output of the virtual power plant; Price(t): electricity market price; C storage (t): operating cost of the energy storage system;

[0116] Apply intelligent optimization algorithms (such as mixed integer programming or genetic algorithms) to dispatch the output power and energy storage charging and discharging behavior of each wind farm; output the dispatch plan after optimization, including the wind farm power generation plan and energy storage equipment operation strategy;

[0117] By integrating wind farms and energy storage resources in the region, virtual power plants can achieve unified optimization of regional power output, enhance the competitiveness of wind power in the power market, and improve the stability and economic benefits of overall power output.

[0118] Please see attached Figure 1 In a preferred embodiment of the present invention, S7 includes:

[0119] Based on the lidar wind speed sequence data and machine learning algorithms, abnormal wind conditions are detected in real time. First, the time series features are extracted from the wind speed v(t), wind direction θ(t) and turbulence intensity τ(t) sequences collected by the lidar in real time. The data is normalized to eliminate dimensional differences and ensure the consistency of feature dimensions.

[0120] Use machine learning algorithms (such as the Isolation Forest algorithm or Support Vector Machine) to build anomaly detection models; feature selection includes wind speed standard deviation, wind speed increment, turbulence change rate, wind direction change rate, etc.;

[0121] Define the anomaly score S(x):

[0122]

[0123] Where, x: sample to be tested; T i : Decision tree; n: number of decision trees,The anomaly detection model can effectively identify abnormal conditions such as turbulence, wind shear or extreme wind conditions by capturing the nonlinear patterns and mutation characteristics in the wind speed sequence, which not only improves the wind farm's ability to respond to extreme wind conditions, but also avoids equipment damage or reduced operating efficiency caused by abnormal wind conditions;

[0124] When turbulence, wind shear or extreme wind conditions are detected, the wind turbine protection mechanism is triggered. Special protection mechanisms are designed for different types of abnormal wind conditions to reduce the impact of abnormal wind conditions on wind turbines and wind farm operations, and to achieve active protection of equipment and operating status, thereby reducing the impact of abnormal wind conditions on equipment, extending the service life of wind turbines, and improving the operational safety of wind farms;

[0125] Combining wind turbine operating data and power deviation, the operating status of the wind turbine is diagnosed and a fault alarm is generated. The status diagnosis model that combines power deviation and operating data can quickly identify potential wind turbine faults and take preventive maintenance measures in advance to avoid further expansion of faults, thereby improving the operation and maintenance efficiency of wind farms, reducing unplanned downtime, and at the same time reducing maintenance costs and improving equipment availability and economic benefits.

[0126] Please see attached Figure 1 In a preferred embodiment of the present invention, the interpolation algorithm is a Kriging interpolation algorithm. During the interpolation process, the interpolation weight is dynamically adjusted according to the terrain characteristics of the wind field, so that the interpolation result is more suitable for the wind speed distribution under complex terrain. First, multi-point wind speed data v is obtained from the real-time measurement of the laser radar. i , the three-dimensional coordinates of the corresponding point are (x i , yi, z i ); Use historical meteorological data and terrain data to supplement background wind speed information outside the coverage area;

[0127] Calculate the spatial covariance between sample points, defined as:

[0128]

[0129] Among them, σ 2 : variance of wind speed variation; φ: spatial correlation parameter; ||xy / / : Euclidean distance between two points;

[0130] At the target point p, the interpolation value is solved by linear weighting:

[0131]

[0132] Where, μ: global mean wind speed; λ i :Interpolation weights are obtained by inverse calculation of the covariance matrix. The Kriging interpolation algorithm can provide high-precision spatial prediction results, and has good adaptability under multi-point measurement and dynamically changing wind field conditions, providing an effective tool for complex terrain wind field modeling;

[0133] Dynamic weight adjustment enhances the adaptability of the interpolation algorithm to complex terrain wind fields. The generated wind speed distribution is more consistent with the actual wind field characteristics, which improves the physical consistency of the model.

[0134] Please see attached Figure 1 In a preferred embodiment of the present invention, short-term wind energy forecasting adopts a deep learning algorithm based on time series. The historical operation data and terrain features of the wind farm are introduced in the model training process to improve the accuracy and real-time performance of the forecast results. The deep learning model can extract implicit feature patterns from complex time series data. Combined with terrain feature data, the model's ability to predict wind speed changes can be further optimized. The multidimensional data introduced in the model construction stage makes wind energy forecasting more comprehensive and accurate, providing high-quality forecasting results for dynamic scheduling and energy storage optimization of wind farms. At the same time, the model optimizes its ability to capture wind speed time series by learning a large amount of historical data, and dynamically adjusts parameters to adapt to the specific wind conditions of different wind farms.

[0135] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A wind farm energy management method based on laser radar wind measurement, characterized in that: The following steps are involved: S1. Deploy lidar equipment to collect real-time data on wind speed, wind direction and turbulence intensity in the wind farm; S2, build a three-dimensional dynamic model of the wind field, combine lidar data and historical meteorological data to model wind speed, wind direction and turbulence distribution; S3. Use deep learning models to make short-term predictions of future wind speeds and output wind energy forecast data from minute to hour levels. S4. Optimize the dynamic dispatch strategy of wind turbines based on forecast data; S5. Connect to the energy storage system, optimize the charging and discharging behavior of the energy storage equipment, and achieve smooth power output; S6. Integrate the operating data of multiple wind farms and optimize regional power synergy through virtual power plants; S7. Detect abnormal wind conditions and diagnose the operating status of wind turbines, and trigger corresponding operating protection or alarm.

2. The wind farm energy management method based on laser radar wind measurement according to claim 1 is characterized in that: The S1 includes: Deploy LiDAR equipment in key areas of the wind farm to ensure that the equipment measurement range covers the main operating areas of the wind farm; Set the LiDAR measurement parameters, including measurement frequency, scanning angle, and measurement range; Collect wind speed, wind direction and turbulence intensity data in real time and transmit the data to the data processing center.

3. The wind farm energy management method based on laser radar wind measurement according to claim 1 is characterized in that: The S2 includes: Fusion of wind speed, wind direction and turbulence intensity data collected in real time by lidar with historical meteorological data; Interpolation algorithms are used to interpolate multi-point data in the wind farm to construct a three-dimensional dynamic model of the wind farm; The turbulence intensity distribution and wake effect in the wind field model are corrected to generate the optimized results of the wind speed distribution in the wind field.

4. The wind farm energy management method based on laser radar wind measurement according to claim 1 is characterized in that: The S3 includes: Extract the time series features of real-time lidar data, including wind speed change trends, wind direction change trends, and turbulence change trends; Short-term wind energy forecasting using deep learning models; Output prediction results for dynamic scheduling and energy storage optimization.

5. The wind farm energy management method based on laser radar wind measurement according to claim 1 is characterized in that: The S4 includes: Dynamically adjust wind turbine operating parameters, including pitch angle and speed ratio; Calculate the optimal value of wind turbine power output using optimization algorithm; Select appropriate wind turbine start and stop strategies under different wind conditions to maximize wind energy utilization.

6. The wind farm energy management method based on laser radar wind measurement according to claim 1 is characterized in that: The S5 includes: Predict the difference between wind power output and grid power demand; Adjust the charging and discharging strategy of the energy storage system to make the power output smooth; Dynamically optimize energy storage power distribution according to the status of the energy storage system and load demand.

7. The wind farm energy management method based on laser radar wind measurement according to claim 1 is characterized in that: The S6 includes: Integrate real-time operating data from multiple wind farms; Construct a regional wind energy collaborative prediction model to predict the wind power output capacity in the region; The power output of regional wind farms is optimized and dispatched through virtual power plants.

8. The wind farm energy management method based on laser radar wind measurement according to claim 1, characterized in that: The S7 includes: Real-time detection of abnormal wind conditions based on lidar wind speed sequence data and machine learning algorithms; Triggering wind turbine protection mechanisms when turbulence, wind shear or extreme wind conditions are detected; Combined with the fan operating data and power deviation, the fan operating status is diagnosed and a fault alarm is generated.

9. The wind farm energy management method based on laser radar wind measurement according to claim 3 is characterized in that: The interpolation algorithm is a Kriging interpolation algorithm. During the interpolation process, the interpolation weight is dynamically adjusted according to the terrain characteristics of the wind field, so that the interpolation result is more suitable for the wind speed distribution under complex terrain.

10. The wind farm energy management method based on laser radar wind measurement according to claim 4, characterized in that: The short-term wind energy forecast adopts a deep learning algorithm based on time series, and the historical operation data and terrain characteristics of the wind farm are introduced into the model training process to improve the accuracy and real-time performance of the forecast results.

Images