Wind power ultra-short-term power prediction method fusing time sequence characteristics and statistical characteristics

Abstract

The invention relates to a wind power ultra-short-term power prediction method fusing time sequence characteristics and statistical characteristics. The invention aims to provide a wind power ultra-short-term power prediction method fusing time sequence characteristics and statistical characteristics. According to the technical scheme, the wind power ultra-short-term power prediction method fusing the time sequence characteristics and the statistical characteristics is characterized by comprising the steps: obtaining and correcting NWP data; inputting the wind speed and wind direction in the corrected NWP data into a CNN network model to extract statistical characteristics; acquiring historical power data, and cleaning the historical power data; inputting the cleaned historical power data into a GRU network model to extract time sequence characteristics; and fusing the statistical characteristics of the wind speed and the wind direction and the time sequence characteristics of the power data, inputting the fused characteristics into the trained ultra-short-term wind power prediction model, and outputting a wind power prediction value. The method is suitable for the field of wind power generation power prediction.

Description

technical field [0001] The invention relates to a wind power ultra-short-term power prediction method that combines time series features and statistical features. It is applicable to the field of wind power generation power prediction. Background technique [0002] Foreign countries have a history of nearly 20 years in the field of wind farm power prediction. More mature researches include Prediktor developed by Ris National Laboratory in Denmark, WPPT (Wind Power Prediction Tool) by Technical University of Denmark, and Previent developed by Olenburg University in Germany. my country has achieved certain results in the field of wind farm power forecasting, most of which focus on the research of statistical model algorithms. [0003] There are very few studies on wind power forecasting using deep learning technology at home and abroad, and such research is in its infancy. Existing research focuses on the use of deep learning algorithms to predict wind speed, and a small num...

AI Technical Summary

Problems solved by technology

However, it is difficult for a single technology to achieve high-precision forecasting, and it is impossible to fully explore the relationship between future wind power output and various factors. Coupled with the influence of noise data, ultra-short-term wind power forecasting has become a multi-variable and highly nonlinear complex problem.

Images