Improving *day-ahead* Solar Irradiance Time Series Forecasting by Leveraging Spatio-Temporal Context (2023-06-01T00:00:00.000000Z)

TL;DR

A deep learning architecture designed to harness spatio-temporal context using satellite data, to attain highly accurate day-ahead time-series forecasting for any given station, with a particular emphasis on forecasting Global Horizontal Irradiance (GHI).

Abstract

Solar power harbors immense potential in mitigating climate change by substantially reducing CO$_{2}$ emissions. Nonetheless, the inherent variability of solar irradiance poses a significant challenge for seamlessly integrating solar power into the electrical grid. While the majority of prior research has centered on employing purely time series-based methodologies for solar forecasting, only a limited number of studies have taken into account factors such as cloud cover or the surrounding physical context. In this paper, we put forth a deep learning architecture designed to harness spatio-temporal context using satellite data, to attain highly accurate \textit{day-ahead} time-series forecasting for any given station, with a particular emphasis on forecasting Global Horizontal Irradiance (GHI). We also suggest a methodology to extract a distribution for each time step prediction, which can serve as a very valuable measure of uncertainty attached to the forecast. When evaluating models, we propose a testing scheme in which we separate particularly difficult examples from easy ones, in order to capture the model performances in crucial situations, which in the case of this study are the days suffering from varying cloudy conditions. Furthermore, we present a new multi-modal dataset gathering satellite imagery over a large zone and time series for solar irradiance and other related physical variables from multiple geographically diverse solar stations. Our approach exhibits robust performance in solar irradiance forecasting, including zero-shot generalization tests at unobserved solar stations, and holds great promise in promoting the effective integration of solar power into the grid.

Authors

References61 items

A Transformer-based multimodal-learning framework using sky images for ultra-short-term solar irradiance forecasting

Prediction Interval Estimation and Deterministic Forecasting Model Using Ground-Based Sky Image

Multi-Scale Message Passing Neural PDE Solvers

Probabilistic Forecasts of Global Horizontal Irradiance for Solar Systems

A Time Series is Worth 64 Words: Long-term Forecasting with Transformers

MAgNet: Mesh Agnostic Neural PDE Solver

TimesNet: Temporal 2D-Variation Modeling for General Time Series Analysis

Fourier Neural Operator with Learned Deformations for PDEs on General Geometries

Less Is More: Fast Multivariate Time Series Forecasting with Light Sampling-oriented MLP Structures

A review of solar forecasting, its dependence on atmospheric sciences and implications for grid integration: Towards carbon neutrality

FiLM-Ensemble: Probabilistic Deep Learning via Feature-wise Linear Modulation

Are Transformers Effective for Time Series Forecasting?

FiLM: Frequency improved Legendre Memory Model for Long-term Time Series Forecasting

Masked Autoencoders As Spatiotemporal Learners

FourCastNet: A Global Data-driven High-resolution Weather Model using Adaptive Fourier Neural Operators

Transformers in Time Series: A Survey

Message Passing Neural PDE Solvers

FEDformer: Frequency Enhanced Decomposed Transformer for Long-term Series Forecasting

IrradianceNet: Spatiotemporal deep learning model for satellite-derived solar irradiance short-term forecasting

Long Expressive Memory for Sequence Modeling

Linking global to regional climate change

Neural Operator: Learning Maps Between Function Spaces

Autoformer: Decomposition Transformers with Auto-Correlation for Long-Term Series Forecasting

RoFormer: Enhanced Transformer with Rotary Position Embedding

RSAM: Robust Self-Attention Based Multi-Horizon Model for Solar Irradiance Forecasting

ViViT: A Video Vision Transformer

Informer: Beyond Efficient Transformer for Long Sequence Time-Series Forecasting

Machine learning for site-adaptation and solar radiation forecasting

An Image is Worth 16x16 Words: Transformers for Image Recognition at Scale

Fourier Neural Operator for Parametric Partial Differential Equations

Short-Term Solar Irradiance Forecasting Using Calibrated Probabilistic Models

Ensemble solar forecasting using data-driven models with probabilistic post-processing through GAMLSS

Benchmark probabilistic solar forecasts: Characteristics and recommendations

Multipole Graph Neural Operator for Parametric Partial Differential Equations

Hybrid Solar Forecasting Method Using Satellite Visible Images and Modified Convolutional Neural Networks

MESHFREEFLOWNET: A Physics-Constrained Deep Continuous Space-Time Super-Resolution Framework

MetNet: A Neural Weather Model for Precipitation Forecasting

Reformer: The Efficient Transformer

A review of deep learning for renewable energy forecasting

Baseline Surface Radiation Network (BSRN): structure and data description (1992–2017)

Decoupled Weight Decay Regularization

Solar Irradiance Forecasting Using Deep Neural Networks

Attention is All you Need

Simple and Scalable Predictive Uncertainty Estimation using Deep Ensembles

Layer Normalization

Validation of models that estimate the clear sky global and beam solar irradiance

Gaussian Error Linear Units (GELUs)

Neural Machine Translation by Jointly Learning to Align and Translate

TV-L1 Optical Flow Estimation

Horizon at station Cabauw

Quantile Regression

Quantile Regression Forests

Crossformer: Transformer Utilizing Cross-Dimension Dependency for Multivariate Time Series Forecasting

pvlib/pvlib-python

Deeponet: Learning nonlinear operators for identifying differential equations based on the universal approximation theorem of operators

Intra-hour direct normal irradiance forecasting through adaptive clear-sky modelling and cloud tracking