PV power forecasting method using a dynamic spatio-temporal attention graph convolutional network with error correction

doi:10.1016/j.solener.2025.113770

PV power forecasting method using a dynamic spatio-temporal attention graph convolutional network with error correction

2025-08-02

0

PRE

AI

DOI：10.1016/j.solener.2025.113770

Original

Original request for help

Save

Abstract

En 中文

• Multi-source NWP contains more information conducive to PV forecasting. • The interpretable method can find the features that really affect the model. • The spatio-temporal dynamic graph overcomes the limitations of the static graph. • The attention mechanism is conducive to forecasting and explaining the model. • The twice decomposition method is used to analyze the internal mechanism of error.

Keywords:

Multi-source NWP

interpretable method

spatio-temporal dynamic graph

attention mechanism

error decomposition

PV power forecasting method using a dynamic spatio-temporal attention graph convolutional network with error correction

Abstract

Journal

Organization