Distributed solar generation forecasting using attention-based deep neural networks for cloud movement prediction

TL;DR

This work proposes an attention-based convolutional long short-term memory network to forecast cloud movement and applies an existing self-attention-based method previously proposed for video prediction to forecast cloud movement and investigates and discusses the impact of cloud forecasts from attention-based methods towards forecasting distributed solar generation, compared to cloud forecasts from non-attention-based methods.

Abstract

Accurate forecasts of distributed solar generation are necessary to reduce negative impacts resulting from the increased uptake of distributed solar photovoltaic (PV) systems. However, the high variability of solar generation over short time intervals (seconds to minutes) caused by cloud movement makes this forecasting task difficult. To address this, using cloud images, which capture the second-to-second changes in cloud cover affecting solar generation, has shown promise. Recently, deep neural networks with "attention" that focus on important regions of an image have been applied with success in many computer vision applications. However, their use for forecasting cloud movement has not yet been extensively explored. In this work, we propose an attention-based convolutional long short-term memory network to forecast cloud movement and apply an existing self-attention-based method previously proposed for video prediction to forecast cloud movement. We investigate and discuss the impact of cloud forecasts from attention-based methods towards forecasting distributed solar generation, compared to cloud forecasts from non-attention-based methods. We further provide insights into the different solar forecast performances that can be achieved for high and low altitude clouds. We find that for clouds at high altitudes, the cloud predictions obtained using attention-based methods result in solar forecast skill score improvements of 5.86% or more compared to non-attention-based methods.

Figures (13)

• Figure 1: Steps involved for two different approaches (direct vs indirect) of forecasting solar generation using satellite or ground-based sky images. Indirect solar forecasting involves a two step process: forecasting cloud movement and using cloud forecasts as an input to the solar forecasting step. Direct solar forecasting approaches bypass the cloud forecasting step and builds a direct mapping between the cloud images and solar generation.
• Figure 2: Locations of the 50 PV sites in Perth and infrared satellite image from Himawari. Locations shown in Figs. \ref{['fig:solar_site_locations']} and \ref{['fig:perth_SI']} are in fact the same, but look different due to different projections.
• Figure 3: Solar power generation time series from a PV site and infrared satellite images corresponding to the time marked in red in the PV power generation time series. The location of the PV site in the satellite image is marked in a red rectangle which indicates the cloud condition vertically above the PV site. Areas without clouds are shown in blue color. Color scale of the satellite images refer to the pixel value that correlates to altitude levels of clouds (1- high, 0-low).
• Figure 4: Solar power generation of a PV site in Perth on eight different days and the pixel value of the infrared satellite image corresponding to the cloud condition above the site.
• Figure 5: Solar power forecasting framework to study the impact of cloud movement predictions from deep neural networks.