Ultra-short-term adjacent wind power prediction method
By scanning the area in front of the wind turbine with lidar, calculating the average wind speed across the turbine cross-section and mapping it to the power generation, second-level wind power prediction is achieved. This solves the problem that existing technologies cannot meet the real-time power balance requirements of off-grid renewable energy microgrids, and improves prediction accuracy and system stability.
Patent Information
- Application Number
- CN202511578889.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-10-31
- Publication Date
- 2026-01-23
AI Technical Summary
Existing wind power forecasting technologies with a 15-minute forecast resolution cannot meet the real-time power balance and frequency stability requirements of off-grid renewable energy microgrids.
By using lidar to scan in front of the wind turbine, wind speed and direction data are obtained through the measurement layer. The average wind speed of the wind turbine cross section is calculated and mapped to the wind turbine's power generation, achieving second-level prediction.
By reducing the minimum time resolution of wind power forecasting technology to the second level, the accuracy of ultra-short-term power forecasting is improved, ensuring real-time power balance and frequency stability of off-grid renewable energy microgrids.
Smart Images
Figure CN121395286A_ABST
Abstract
Description
TECHNICAL FIELD
[0001] The present application relates to the field of wind power prediction technology, and in particular to a method for ultra-short-term near-wind power prediction. BACKGROUND
[0002] At present, the proportion of new energy represented by wind power in the energy system structure is increasing. In order to solve the problem of power system power balance caused by the volatility, intermittency and randomness of wind power, the method of predicting wind power is usually used to predict the change of wind power in advance, so as to deploy other power or load for adjustment, so as to ensure that the total power generation of the power system matches the load demand, maintain the power balance of the power system, and ensure the stable operation of the power system.
[0003] In grid-connected wind power projects, the minimum time resolution of the mainstream and mature wind power prediction technology in the industry is 15 minutes. However, in off-grid new energy microgrids, due to the lack of large grid support, it is necessary to maintain power balance and frequency stability in real time, so that flexible load can quickly track the changes of front-end wind power. However, the existing 15-minute level prediction resolution cannot meet the requirements of real-time power balance and frequency stability of off-grid new energy microgrids. SUMMARY
[0004] In order to reduce the time resolution of wind power prediction and improve the accuracy of wind power prediction, the present application provides a method for ultra-short-term near-wind power prediction.
[0005] The method for ultra-short-term near-wind power prediction provided by the present application adopts the following technical scheme: A method for ultra-short-term near-wind power prediction, comprising the following steps: Step S1, dividing the wind turbine cross section into measuring zones, and dividing the front of the wind turbine into measuring layers, wherein and are positive integers; Step S2, scanning the front of the wind turbine by laser radar; Step S3, obtaining the wind speed data of the th measuring layer at the th measuring zone at the second is a positive integer and , wherein is a positive integer and ; obtaining the wind direction data of the th measuring layer at the th measuring zone at the second is a positive integer and , wherein is a positive integer and ; Step S4, calculating the future second moment fan cross-section average wind speed ; Step S5, according to the future second moment fan cross-section average wind speed mapping, the future second moment fan power generation power is obtained.
[0006] By adopting the above technical scheme, the laser radar installed on the fan scans the air in front of the fan by using the laser beam, the signal frequency changes after the laser beam interacts with the aerosol or dust particles, so as to obtain the wind speed data and wind direction data of different measurement layers at different measurement areas, when predicting the wind speed at the future moment, the wind speed average value is calculated according to the wind speed data in the past time period, the wind direction average value is calculated according to the wind direction data in the past time period, the horizontal wind speed component of each position is obtained by using the wind speed average value and the wind direction average value, then the weighted average of the horizontal wind speed component of each position is carried out, so as to obtain the future moment fan cross-section average wind speed, after obtaining the future moment fan cross-section average wind speed, the corresponding fan power generation power is obtained through the wind speed power mapping model, so as to predict the future moment fan power generation power, the minimum time resolution of wind power prediction technology is reduced from 15 minutes to seconds, and the accuracy of the ultra-short-term power prediction is improved.
[0007] As a preferred, the measurement layers are all arranged in parallel with the fan cross-section, and the center points of the measurement layers all coincide with the emission end of the laser radar.
[0008] By adopting the above technical scheme, the projection planes of the measurement layers all coincide, and the emission end of the laser radar is located at the center of the projection plane of the measurement layer, so that the laser beams emitted by the laser radar are uniformly distributed on the measurement layer.
[0009] As a preferred, the adjacent measurement layers are spaced apart by 20m, and the adjacent measurement layers are spaced apart from the fan by 20m.
[0010] By adopting the above technical scheme, it is convenient to obtain the wind speed data and wind direction data which are uniformly distributed.
[0011] As a preferred, the laser radar emits the first laser beam, the second laser beam, the third laser beam and the fourth laser beam.
[0012] Preferably, the first laser beam has an angle of 30° with the horizontal plane, an angle of 25° with the vertical plane, the second laser beam has an angle of 30° with the horizontal plane, an angle of -25° with the vertical plane, the third laser beam has an angle of -30° with the horizontal plane, an angle of 25° with the vertical plane, and the fourth laser beam has an angle of -30° with the horizontal plane and an angle of -25° with the vertical plane.
[0013] By adopting the above technical solution, the first laser beam, the second laser beam, the third laser beam and the fourth laser beam are emitted from the transmitter of the lidar in all directions, which facilitates the uniform distribution of the laser beam emitted by the lidar on the measurement layer.
[0014] Preferably, step S4 includes the following steps: Step S41: Calculate the future Seconds The measurement layer is at the first Predicted wind speed at the measurement area ; Calculating the future Seconds The measurement layer is at the first Predicted wind direction at the measurement area ; Step S42: Calculate the future Seconds The measurement layer is at the first Predicted horizontal wind speed components in the measurement area , ; Step S43: Calculate the future Average wind speed at the fan cross section at a given time (second) , ,in It is the first The area of the measurement zone.
[0015] By adopting the above technical solution, the average wind speed and wind direction data of each measurement layer in each measurement area are obtained based on the wind speed data and wind direction data of the corresponding location in the past period. The average wind speed and wind direction data of the corresponding location in the past period are calculated. Then, the horizontal wind speed component of the location is obtained by using the average wind speed and wind direction data of the location. Finally, the horizontal wind speed components of each location are weighted and averaged to obtain the average wind speed of the wind turbine cross section at future time.
[0016] Preferably, in step S41 , It is the window size and , Is The wind speed data measured at the moment.
[0017] By adopting the technical scheme, the wind speed average value is calculated according to the wind speed data in the past time period of the corresponding position.
[0018] As preferred, in the step S41 , is the window size and , is in The wind direction data measured at the moment.
[0019] By adopting the technical scheme, the wind direction average value is calculated according to the wind direction data in the past time period of the corresponding position.
[0020] In summary, the present application includes at least one of the following beneficial technical effects: 1. According to the wind condition information provided by the laser radar, the wind power can be predicted 5-10 seconds in advance, the minimum time resolution of the wind power prediction technology is reduced from 15 minutes to seconds, and the accuracy of the ultra-short-term power prediction is improved; 2. According to the wind power predicted 5-10 seconds in advance, the rear-end flexible load is adjusted, which is convenient for meeting the requirements of off-grid new energy micro-grid real-time power balance and frequency stability, and is helpful for safe and stable operation of the wind power system. BRIEF DESCRIPTION OF DRAWINGS
[0021] Figure 1 is a flow chart of an ultra-short-term near wind power prediction method in the embodiment 1 of the present application.
[0022] Figure 2 is a schematic diagram of the laser radar wind measurement in the embodiment 1 of the present application.
[0023] Figure 3 is a wind measurement data distribution diagram of the laser radar in the embodiment 1 of the present application.
[0024] Figure 4 is a flow chart of obtaining the wind turbine power at the future moment in the embodiment 2 of the present application.
[0025] Figure 5 is a wind turbine power curve diagram in the embodiment 2 of the present application.
[0026] Figure 6 is a flow chart of obtaining the wind turbine power at the future moment in the embodiment 3 of the present application.
[0027] Figure 7 is a Simulink wind turbine mechanism model diagram in the embodiment 3 of the present application.
[0028] Figure 8This is a flowchart of obtaining the future wind turbine power generation capacity in Embodiment 4 of this application.
[0029] Figure 9 This is a schematic diagram of the wind turbine power generation prediction model in Embodiment 4 of this application.
[0030] Figure 10 This is a diagram showing the effect of load tracking without advance control in the electrolytic cell in Embodiment 5 of this application.
[0031] Figure 11 This is a diagram illustrating the effect of advanced load tracking control in the electrolytic cell of Embodiment 6 of this application. Detailed Implementation
[0032] The following is in conjunction with the appendix Figures 1-11 This application will be described in further detail.
[0033] Embodiment 1 of this application discloses a method for predicting ultra-short-term near-term wind power. (Refer to...) Figures 1 to 3 This includes the following steps.
[0034] Step S1, the fan cross-section is divided into The measurement area is divided in front of the wind turbine. One measurement layer, in which and All values are positive integers. The measurement layers are all set parallel to the wind turbine cross-section, and their projected surfaces coincide. Adjacent measurement layers are spaced 20m apart; the measurement layer closest to the wind turbine is also spaced 20m apart, while the measurement layer furthest from the wind turbine is spaced 200m apart.
[0035] Step S2: The lidar on the wind turbine emits a first laser beam, a second laser beam, a third laser beam, and a fourth laser beam, with the center point of the measurement layer coinciding with the lidar's emitting end. The first laser beam has an angle of 30° with the horizontal plane and 25° with the vertical plane; the second laser beam has an angle of 30° with the horizontal plane and -25° with the vertical plane; the third laser beam has an angle of -30° with the horizontal plane and 25° with the vertical plane; and the fourth laser beam has an angle of -30° with the horizontal plane and -25° with the vertical plane. The first, second, third, and fourth laser beams are emitted and scan in front of the wind turbine with the lidar's emitting end as the origin, obtaining uniformly distributed wind speed and direction data.
[0036] Step S3: Verify the wind speed and direction data acquired by the lidar, removing obviously erroneous or abnormal data points. Then, denoise and filter the verified wind speed and direction data. Seconds The measurement layer is at the first Wind speed data at the measurement zones wherein is a positive integer and wherein is a positive integer and ; the laser radar obtains the predicted wind speed of the th measurement layer at the th measurement zone at the future second time point wherein is a positive integer and wherein is a positive integer and .
[0037] Step S4, calculating the future second time point fan cross-section average wind speed , comprising the following steps.
[0038] Step S41, calculating the future second time point predicted wind speed of the th measurement layer at the th measurement zone , , is a window size and , is the wind speed data measured at the second time point; calculating the future second time point predicted wind direction of the th measurement layer at the th measurement zone , , is a window size and , is the wind direction data measured at the second time point; Step S42, calculating the future second time point predicted horizontal wind speed component of the th measurement layer at the th measurement zone , ; Step S43, calculating the future second time point fan cross-section average wind speed , wherein is the area of the th measurement zone.
[0039] Step S5, mapping the future second time point fan cross-section average wind speed to obtain the future second time point fan power generation power.
[0040] The implementation principle of the ultra-short-term near-term wind power prediction method in this application is as follows: A lidar installed on a wind turbine scans the air in front of the turbine using a laser beam. After the laser beam interacts with aerosols or dust particles, the frequency of the returned signal changes, thereby obtaining wind speed and direction data at different measurement layers and in different measurement areas. When predicting the wind speed at a future time, the average wind speed is calculated based on the wind speed data of the corresponding location over a certain period of time, and the average wind direction is calculated based on the wind direction data of the corresponding location over a certain period of time. Then, the horizontal wind speed component at that location is obtained using the average wind speed and average wind direction data. Finally, the horizontal wind speed components at each location are weighted and averaged to obtain the average wind speed of the turbine cross-section at the future time. After obtaining the average wind speed of the turbine cross-section at the future time, the corresponding wind turbine power generation can be obtained through a wind speed-power mapping model, thereby predicting the wind turbine power generation at the future time. Based on the wind condition information provided by the lidar, wind power can be predicted 5-10 seconds in advance, reducing the minimum time resolution of wind power prediction technology from 15 minutes to the second level, thus improving the accuracy of ultra-short-term power prediction. Based on the wind power predicted 5-10 seconds in advance, the flexible load at the back end is adjusted to meet the requirements of off-grid new energy microgrids to maintain power balance and frequency stability in real time, which helps the wind power system to operate safely and stably.
[0041] Embodiment 2 of this application discloses a method based on the future Average wind speed at the fan cross section at a given time (second) Mapping to obtain the future Method for determining wind turbine power generation per second. (Refer to...) Figure 4 and Figure 5 According to the future Average wind speed at the fan cross section at a given time (second) By combining the power generation curve of the wind turbine, the power of the wind turbine at the corresponding wind speed can be obtained, thus predicting the future power output. The wind turbine's power generation capacity at any given second. The power generation curve of this wind turbine is obtained by fitting data from the wind turbine supplier or actual wind turbine operating data.
[0042] Embodiment 3 of this application discloses a method based on the future Average wind speed at the fan cross section at a given time (second) Mapping to obtain the future Method for determining wind turbine power generation per second. (Refer to...) Figure 6 and Figure 7 The wind farm data in front of the wind turbine and the future Average wind speed at the fan cross section at a given time (second) The wind turbine's operating mechanism model is fed into the input data, and the power output at the corresponding wind speed is calculated through simulation using the model, thus achieving the effect of advanced power prediction. The wind turbine operating mechanism model is obtained from wind turbine suppliers or from commonly used wind turbine models available in simulation software such as Simulink.
[0043] Embodiment 4 of this application discloses a method based on the future Average wind speed at the fan cross section at a given time (second) Mapping to obtain the future Method for determining wind turbine power generation per second. (Refer to...) Figure 8 and Figure 9 The actual wind measurement data and wind turbine power generation data are processed and feature-engineered before being input into different machine learning models for training. Based on the prediction accuracy of the trained models, an appropriate model is selected for power prediction, or a federated model is used to assign different weights to the prediction results of different models, ultimately outputting the wind turbine power prediction result. After a certain period, newly accumulated data needs to be updated into the training dataset, and the model is retrained, continuously performing this optimization iteration process. Machine learning models can be obtained from open-source models such as SVM, XGBoost, and LSTM, or they can be obtained by building a multi-layer neural network model or a deep learning model.
[0044] Embodiment 5 of this application discloses a simulation method for load tracking in an electrolytic cell without advance control. (Refer to...) Figure 10 When the electrolytic cell adjustment speed is only 0.2% / s and there is no advance control, the maximum power deviation is 38MW, the total energy gap is 6.8MWh, the total electricity consumption is 3533MWh, the energy storage configuration requires 38MW / 38MWh, the grid connection requires a transmission channel capacity of not less than 38MW, and the connection rate is about 6.8 / 3533=0.2%.
[0045] Embodiment 6 of this application discloses a simulation method for load tracking in an electrolytic cell without advance control. (Refer to...) Figure 11 When the electrolyzer regulation rate is only 0.2% / s, the load tracking capability improves with the addition of predictive control. The maximum power deviation is 19MW, the total energy gap is 3.1MWh, and the total electricity consumption is 3533MWh. Relying on energy storage requires 19MW / 19MWh, while relying on grid power requires a transmission channel capacity of no less than 19MW, with a power-down rate of approximately 3.1 / 3533 = 0.09%. The addition of predictive control significantly improves the load tracking capability. Analysis of the calculation results shows that by issuing control commands to the electrolyzer in advance through near-term predictive power, eliminating the impact of communication delays, the required energy storage configuration is reduced from 38MW / 38MWh to 19MW / 19MWh, reducing energy storage costs by approximately 19 million RMB.
[0046] The above are all preferred embodiments of the present application, and do not limit the protection scope of the present application, so that: all equivalent changes made according to the structure, shape, principle of the present application should be covered in the protection scope of the present application.
Claims
1. A method for predicting ultra-short-term near-term wind power, characterized in that: Includes the following steps: Step S1, the fan cross-section is divided into The measurement area is divided in front of the wind turbine. One measurement layer, in which and All are positive integers; Step S2: Use lidar to scan the area in front of the fan; Step S3, Obtain Seconds The measurement layer is at the first Wind speed data at the measurement area ,in are positive integers and ,in are positive integers and ;get Seconds The measurement layer is at the first Wind direction data at the measurement area ,in are positive integers and ,in are positive integers and ; Step S4: Calculate the future Average wind speed at the fan cross section at a given time (second) ; Step S5, based on the future Average wind speed at the fan cross section at a given time (second) Mapping to obtain the future The wind turbine's power generation capacity per second.
2. The method for predicting ultra-short-term near-term wind power according to claim 1, characterized in that: The measurement layers are all arranged parallel to the cross-section of the wind turbine, and the center point of each measurement layer coincides with the transmitting end of the lidar.
3. The method for predicting ultra-short-term near-term wind power according to claim 1, characterized in that: The adjacent measurement layers are spaced 20m apart, and the adjacent measurement layers are spaced 20m apart from the fan.
4. The method for predicting ultra-short-term near-term wind power according to claim 1, characterized in that: The lidar emits a first laser beam, a second laser beam, a third laser beam, and a fourth laser beam.
5. The method for predicting ultra-short-term near-term wind power according to claim 4, characterized in that: The first laser beam has an angle of 30° with the horizontal plane and an angle of 25° with the vertical plane. The second laser beam has an angle of 30° with the horizontal plane and an angle of -25° with the vertical plane. The third laser beam has an angle of -30° with the horizontal plane and an angle of 25° with the vertical plane. The fourth laser beam has an angle of -30° with the horizontal plane and an angle of -25° with the vertical plane.
6. The method for predicting ultra-short-term near-term wind power according to claim 1, characterized in that: Step S4 includes the following steps: Step S41: Calculate the future Seconds The measurement layer is at the first Predicted wind speed at the measurement area ; Calculating the future Seconds The measurement layer is at the first Predicted wind direction at the measurement area ; Step S42: Calculate the future Seconds The measurement layer is at the first Predicted horizontal wind speed components in the measurement area , ; Step S43: Calculate the future Average wind speed at the fan cross section at a given time (second) , ,in It is the first The area of the measurement zone.
7. The method for predicting ultra-short-term near-term wind power according to claim 6, characterized in that: In step S41 , It is the window size and , Is Wind speed data measured at all times.
8. The method for predicting ultra-short-term near-term wind power according to claim 6, characterized in that: In step S41 , It is the window size and , Is Wind direction data measured at all times.