Open-source sky image datasets for solar forecasting with deep learning: A comprehensive survey

Nie, Yuhao; Li, Xiatong; Paletta, Quentin; Aragon, Max; Scott, Andea; Brandt, Adam

doi:10.1016/j.rser.2023.113977

Cited by 19 publications

(3 citation statements)

References 168 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are also more locally focused weather predictions available, such as the Royal Netherlands Meteorological Institute (KNMI) for the Netherlands. And when looking for ASI images, an extensive list of 72 open-source, sky image datasets was put together by Nie et al [102]. This list also details whether or not the images are complemented with additional data, such as solar irradiance or the PV power output, and whether they are selected to be suitable for AI-driven methods.…”

Section: Creation Of Publicly Available Standardized Datasetsmentioning

confidence: 99%

A Review of Solar Forecasting Techniques and the Role of Artificial Intelligence

Barhmi,

Heynen,

Golroodbari

et al. 2024

Solar

View full text Add to dashboard Cite

Solar energy forecasting is essential for the effective integration of solar power into electricity grids and the optimal management of renewable energy resources. Distinguishing itself from the existing literature, this review study provides a nuanced contribution by centering on advancements in forecasting techniques. While preceding reviews have examined factors such as meteorological input parameters, time horizons, the preprocessing methodology, optimization, and sample size, our study uniquely delves into a diverse spectrum of time horizons, spanning ultrashort intervals (1 min to 1 h) to more extended durations (up to 24 h). This temporal diversity equips decision makers in the renewable energy sector with tools for enhanced resource allocation and refined operational planning. Our investigation highlights the prominence of Artificial Intelligence (AI) techniques, specifically focusing on Neural Networks in solar energy forecasting, and we review supervised learning, regression, ensembles, and physics-based methods. This showcases a multifaceted approach to address the intricate challenges associated with solar energy predictions. The integration of Satellite Imagery, weather predictions, and historical data further augments precision in forecasting. In assessing forecasting models, our study describes various error metrics. While the existing literature discusses the importance of metrics, our emphasis lies on the significance of standardized datasets and benchmark methods to ensure accurate evaluations and facilitate meaningful comparisons with naive forecasts. This study stands as a significant advancement in the field, fostering the development of accurate models crucial for effective renewable energy planning and emphasizing the imperative for standardization, thus addressing key gaps in the existing research landscape.

show abstract

Section: Creation Of Publicly Available Standardized Datasetsmentioning

confidence: 99%

A Review of Solar Forecasting Techniques and the Role of Artificial Intelligence

Barhmi,

Heynen,

Golroodbari

et al. 2024

Solar

View full text Add to dashboard Cite

show abstract

“…Image datasets (Nie et al, 2024), (Pawar & Ainapure, 2023) encompass a wide range of image types and serve as collections of digital data. The image dataset is sourced from Kaggle, Google Images, and GitHub repositories.…”

Section: Datasetmentioning

confidence: 99%

Revolution in Image Data Collection: CycleGAN as a Dataset Generator

Hindarto,

Handayani

2024

SinkrOn

View full text Add to dashboard Cite

Computer vision, deep learning, and pattern recognition are just a few fields where image data collection has become crucial. The Cycle Generative Adversarial Network has become one of the most effective instruments in the recent revolution in image data collection. This research aims to comprehend the impact of CycleGAN on the collection of image datasets. CycleGAN, a variant of the Generative Adversarial Network model, has enabled the unprecedented generation of image datasets. CycleGAN can transform images from one domain to another without manual annotation by employing adversarial learning between the generator and discriminator. This means generating image datasets quickly and efficiently for various purposes, from object recognition to data augmentation. One of the most fascinating features of CycleGAN is its capacity to alter an image's style and characteristics. Using CycleGAN to generate unique and diverse datasets assists deep learning models in overcoming visual style differences. This is a significant development in understanding how machine learning models can comprehend visual art concepts. CycleGAN's use as a data set generator has altered the landscape of image data collection. CycleGAN has opened new doors in technological innovation and data science with its proficiency in generating diverse and unique datasets. This research will investigate in greater detail how CycleGAN revolutionized the collection of image datasets and inspired previously unconceived applications.

show abstract

“…However, the assimilation of solar power into the grid poses distinct challenges, primarily because of its variable nature. Solar power generation (SPG) is subject to natural phenomena, such as sunlight intensity and duration, causing daily and seasonal fluctuations [3]. Consequently, the precise forecasting of SPG has become vital for maintaining grid stability and optimizing the use of solar energy.…”

Section: Introduction 1motivationmentioning

confidence: 99%

Deep-Reinforcement-Learning-Based Vehicle-to-Grid Operation Strategies for Managing Solar Power Generation Forecast Errors

Jang,

2024

Sustainability

View full text Add to dashboard Cite

This study proposes a deep-reinforcement-learning (DRL)-based vehicle-to-grid (V2G) operation strategy that focuses on the dynamic integration of charging station (CS) status to refine solar power generation (SPG) forecasts. To address the variability in solar energy and CS status, this study proposes a novel approach by formulating the V2G operation as a Markov decision process and leveraging DRL to adaptively manage SPG forecast errors. Utilizing real-world data from the Korea Southern Power Corporation, the effectiveness of this strategy in enhancing SPG forecasts is proven using the PyTorch framework. The results demonstrate a significant reduction in the mean squared error by 40% to 56% compared to scenarios without V2G. Our investigation into the effects of blocking probability thresholds and discount factors revealed insights into the optimal V2G system performance, suggesting a balance between immediate operational needs and long-term strategic objectives. The findings highlight the possibility of using DRL-based strategies to achieve more reliable and efficient renewable energy integration in power grids, marking a significant step forward in smart grid optimization.

show abstract

Open-source sky image datasets for solar forecasting with deep learning: A comprehensive survey

Cited by 19 publications

References 168 publications

A Review of Solar Forecasting Techniques and the Role of Artificial Intelligence

A Review of Solar Forecasting Techniques and the Role of Artificial Intelligence

Revolution in Image Data Collection: CycleGAN as a Dataset Generator

Deep-Reinforcement-Learning-Based Vehicle-to-Grid Operation Strategies for Managing Solar Power Generation Forecast Errors

Contact Info

Product

Resources

About