CN-121983949-A - Second-level wind power prediction method and system based on laser radar data

CN121983949ACN 121983949 ACN121983949 ACN 121983949ACN-121983949-A

Abstract

The invention discloses a second-level wind power prediction method and system based on laser radar data, and relates to the technical field of wind power prediction. And acquiring multisource heterogeneous data of the network, the fan SCADA system and the numerical weather forecast, and obtaining revised data through space correlation anomaly identification and reconstruction. And predicting the total power of the wind power plant by using a power prediction model constructed by the long-short-term memory network, and updating power prediction data and displaying the terminal according to a preset period. The method solves the technical problems that the traditional wind power prediction means is difficult to comprehensively capture key airflow information of the wind power plant, effectively process multisource data anomalies and the prediction response speed cannot be matched with real-time requirements, achieves the technical effects of accurately predicting wind power in real time and providing effective support for efficient scheduling of the wind power plant and stable operation of a power grid.

Inventors

XU CHUN
YU BOWEN
ZHAO ZONGKAI
LIU BING
LI HAIBIN
LIU YAN
GUO WEI
HE WEI
YE JIANFENG
CAO GUOAN

Assignees

大唐(内蒙古)能源开发有限公司
大唐多伦瑞源新能源有限公司
中国大唐集团科技创新有限公司
中新能化科技有限公司
大唐内蒙古多伦煤化工有限责任公司

Dates

Publication Date: 20260505
Application Date: 20251209

Claims (8)

1. The second-level wind power prediction method based on laser radar data is characterized by comprising the following steps of: Deploying a real-time wind measuring network at a key airflow position of a wind farm, wherein the real-time wind measuring network consists of a foundation type wind measuring radar and a cabin type laser wind measuring radar; Acquiring multi-source heterogeneous data, and executing abnormal data identification and reconstruction based on spatial correlation to determine revised data, wherein the multi-source heterogeneous data are from the real-time anemometry network, a fan SCADA system and a numerical weather forecast; Carrying out wind farm total power prediction on the revised data according to a power prediction model, and determining power prediction data, wherein the power prediction model is constructed based on a long-short-time memory network; and according to a preset period, updating the power prediction data and displaying the power prediction data by the terminal.
2. The method for predicting second-level wind power based on laser radar data according to claim 1, wherein deploying a real-time wind network comprises: deploying a foundation type laser wind measuring radar on the top of a prefabricated cabin of a secondary system in the center of a wind power plant; three cabin type laser wind measuring radars are respectively deployed on an upstream fan, a downstream fan and a representative fan around a field area in the main wind direction of the wind power plant, wherein the cabin type laser wind measuring radars are used for capturing inflow, wake flow and boundary airflow information.
3. The method for predicting second-level wind power based on lidar data according to claim 1, wherein performing anomaly data identification and reconstruction based on spatial correlation, determining revised data, comprises: aiming at each measuring point in a wind power plant, establishing a spatial correlation model by quantifying the spatial correlation among wind speed data of each measuring point; The space correlation model reads the multi-source heterogeneous data and identifies abnormal data points; performing reconstruction on the abnormal data points to determine reconstructed data points; and replacing abnormal data points in the multi-source heterogeneous data with reconstructed data points, and determining the revised data.
4. The method for predicting second-level wind power based on laser radar data according to claim 3, wherein the reconstruction mode of the abnormal data points is that M points with optimal data quality are selected from K normal measuring points with highest correlation with the abnormal data points, spatial interpolation is carried out based on spatial correlation, and the reconstructed data points are generated, wherein M is a positive integer less than or equal to K.
5. The method for predicting the second-level wind power based on the laser radar data according to claim 4, wherein a dynamic trend consistency test is introduced as an auxiliary criterion for data anomaly; When any data point is detected to jump, and the change mode is inconsistent with the historical trend of the data point in the sliding time window, the data point is marked as a suspected abnormal point to trigger early warning even if the data deviation threshold is not exceeded.
6. The method for predicting wind power in second level based on laser radar data according to claim 1, wherein the constructing the power prediction model before predicting the total power of the wind farm for the revised data according to the power prediction model comprises: Acquiring training data, wherein the training data comprises historical data samples, and each sample sequence comprises laser radar data, fan operation data, numerical weather forecast data and actual power data; and taking the long-term memory network as a core network structure, performing supervision training to converge according to the training data, and generating the power prediction model.
7. The method for predicting second-level wind power based on laser radar data according to claim 6, wherein the output mode of the power prediction model is to take a preset time period as a prediction time zone and take a preset prediction point interval and a preset prediction point number in the preset time period as prediction data amounts.
8. A second-level wind power prediction system based on laser radar data, for implementing the second-level wind power prediction method based on laser radar data according to any one of claims 1 to 7, the system comprising: the real-time wind measuring network deployment module is used for deploying a real-time wind measuring network at a key airflow position of a wind power plant, wherein the real-time wind measuring network consists of a foundation type wind measuring radar and a cabin type laser wind measuring radar; The revision data acquisition module is used for acquiring multi-source heterogeneous data, executing abnormal data identification and reconstruction based on spatial correlation, and determining revision data, wherein the multi-source heterogeneous data are from the real-time anemometry network, a fan SCADA system and a numerical weather forecast; The power prediction data acquisition module is used for predicting the total power of the wind power plant according to the revised data and determining power prediction data, wherein the power prediction model is constructed based on a long-short-time memory network; And the terminal display execution module is used for updating and displaying the power prediction data according to a preset period.

Description

Second-level wind power prediction method and system based on laser radar data Technical Field The invention relates to the technical field of wind power prediction, in particular to a second-level wind power prediction method and system based on laser radar data. Background Wind power prediction is a key for guaranteeing efficient dispatching of wind power plants and stable grid connection of power grids, and second-level prediction is particularly important for real-time management and control. The prior art relies on fan SCADA data, numerical weather forecast and conventional wind measuring equipment, and plays a role in a simple scene. However, as the wind power control requirements are improved, the traditional means are exposed and limited in that the key airflow area of the wind power plant is complex, the core airflow points cannot be covered by the conventional wind measurement, and the exception of the multi-source data is easy to exist and is not effectively processed. The traditional technology can not accurately acquire the comprehensive airflow and effective data of the wind power plant, so that the power prediction is inaccurate and the real-time performance is insufficient, and the requirements of second-level wind power accurate prediction and effective management and control of the wind power plant are difficult to meet. Disclosure of Invention The application provides a second-level wind power prediction method and a second-level wind power prediction system based on laser radar data, which solve the technical problems that the traditional wind power prediction means is difficult to comprehensively capture key airflow information of a wind power plant, multi-source data abnormality is effectively processed, and the prediction response speed cannot be matched with real-time requirements. The application provides a second-level wind power prediction method based on laser radar data, which comprises the steps of deploying a real-time wind measuring network at a key airflow position of a wind power plant, wherein the real-time wind measuring network consists of a foundation-type wind measuring radar and a cabin-type laser wind measuring radar, collecting multi-source heterogeneous data, executing abnormal data identification and reconstruction based on spatial correlation, determining revised data, wherein the multi-source heterogeneous data are from the real-time wind measuring network, a fan SCADA system and numerical weather forecast, performing wind power plant total power prediction on the revised data according to a power prediction model, determining power prediction data, constructing the power prediction model based on a long-short time memory network, and executing updating and terminal display of the power prediction data according to a preset period. The second aspect of the application provides a second-level wind power prediction system based on laser radar data, which comprises a real-time wind network deployment module, a revision data acquisition module, a power prediction data acquisition module and a terminal display execution module, wherein the real-time wind network deployment module is used for deploying a real-time wind network at a key airflow position of a wind power plant, the real-time wind network consists of a foundation-based wind radar and a cabin-type laser wind radar, the revision data acquisition module is used for acquiring multi-source heterogeneous data, executing abnormal data identification and reconstruction based on space correlation and determining revision data, the multi-source heterogeneous data are from the real-time wind network, a fan SCADA system and numerical weather forecast, the power prediction data acquisition module is used for carrying out total power prediction of the wind power plant according to a power prediction model, the power prediction model is constructed based on a long-short-time memory network, and the terminal display execution module is used for executing updating and terminal display of the power prediction data according to a preset period. One or more technical schemes provided by the application have at least the following technical effects or advantages: according to the method, the wind measuring network is deployed at the key airflow position of the wind power plant to acquire multi-source data, the revised data is acquired through space correlation anomaly identification and reconstruction processing, the total power prediction data of the wind power plant is calculated based on the revised data, and the prediction data is updated and displayed by a terminal according to a preset period, so that the real-time prediction of wind power is accurately realized, the wind power prediction result is more accurate and timely in response, the requirements of accurate control of the wind power plant and stable grid connection of the power grid are met, the real-time accurate prediction of wind power is achieved, and the tec