Spatio-temporal graph neural networks for multi-site PV power forecasting

TL;DR

This paper introduces two novel graph neural network models, GCLSTM and GCTrafo, that leverage spatio-temporal dependencies in PV production data to improve multi-site solar power forecasting accuracy without relying on numerical weather predictions.

Contribution

The paper presents two innovative graph neural network models for PV forecasting that outperform existing methods in multi-site and single-site scenarios using only production data.

Findings

Outperform state-of-the-art multi-site forecasting methods for 6-hour horizons.

Outperform single-site methods with NWP inputs for up to 4-hour horizons.

Effective modeling of spatio-temporal dependencies in PV data.

Abstract

Accurate forecasting of solar power generation with fine temporal and spatial resolution is vital for the operation of the power grid. However, state-of-the-art approaches that combine machine learning with numerical weather predictions (NWP) have coarse resolution. In this paper, we take a graph signal processing perspective and model multi-site photovoltaic (PV) production time series as signals on a graph to capture their spatio-temporal dependencies and achieve higher spatial and temporal resolution forecasts. We present two novel graph neural network models for deterministic multi-site PV forecasting dubbed the graph-convolutional long short term memory (GCLSTM) and the graph-convolutional transformer (GCTrafo) models. These methods rely solely on production data and exploit the intuition that PV systems provide a dense network of virtual weather stations. The proposed methods were evaluated in two data sets for an entire year: 1) production data from 304 real PV systems, and 2) simulated production of 1000 PV systems, both distributed over Switzerland. The proposed models outperform state-of-the-art multi-site forecasting methods for prediction horizons of six hours ahead. Furthermore, the proposed models outperform state-of-the-art single-site methods with NWP as inputs on horizons up to four hours ahead.