Photovoltaic power interval prediction method based on boundary approximation after feature selection

Abstract

The invention discloses a photovoltaic power interval prediction method based on boundary approximation after feature selection. The method comprises the steps: carrying out feature reconstruction of historical weather features, and obtaining recombined features; performing feature selection on the recombined features to obtain screening features; performing photovoltaic power prediction by using the screening features and historical power data to obtain an initial photovoltaic power prediction interval; and performing boundary approximation on the initial prediction interval to obtain a photovoltaic power prediction interval. When the interval coverage rate under different confidence intervals meets confidence requirements, the predicted power interval is optimized to reduce the interval width; the sliding time window weighting based on information entropy can avoid the influence of subjective factors, and the credibility and accuracy of analysis are improved; the problem that the interval width of a part of time periods is too large due to sudden change of weather can be corrected by sliding time window weighting.

Description

technical field [0001] The invention belongs to the technical field of photovoltaic power prediction methods, and relates to a photovoltaic power interval prediction method based on boundary approximation after feature selection. Background technique [0002] Interval forecasting has been widely used in the field of power forecasting because of its ability to characterize uncertainty and avoid the problem of point forecasting forecast accuracy. Most of the existing power interval predictions satisfy the interval coverage, but ignore the interval width, which leads to the problem that the interval width is too large while the interval coverage is high. Since the interval coverage and interval width are contradictory, increasing the interval coverage will increase the interval width. The currently commonly used error correction interval width is only to reduce the error of the model and the calculation process. For the interval width that ensures high interval coverage, the c...

AI Technical Summary

Problems solved by technology

[0003] The purpose of the present invention is to provide a photovoltaic power interval prediction method based on boundary approximation after feature selection, which solves the problem of high interval coverage and excessive interval width existing in the prior art

Images