Machine Learning and Deep Learning in Energy Systems: A Review

Sustainability, Journal Year: 2022, Volume and Issue: 14(8), P. 4832 - 4832

Published: April 18, 2022

With population increases and a vital need for energy, energy systems play an important decisive role in all of the sectors society. To accelerate process improve methods responding to this increase demand, use models algorithms based on artificial intelligence has become common mandatory. In present study, comprehensive detailed study been conducted applications Machine Learning (ML) Deep (DL), which are newest most practical Artificial Intelligence (AI) systems. It should be noted that due development DL algorithms, usually more accurate less error, these ability model solve complex problems field. article, we have tried examine very powerful problem solving but received attention other studies, such as RNN, ANFIS, RBN, DBN, WNN, so on. This research uses knowledge discovery databases understand ML systems’ current status future. Subsequently, critical areas gaps identified. addition, covers efficient used field; optimization, forecasting, fault detection, investigated. Attempts also made cover their evaluation metrics, including not only important, newer ones attention.

Language: Английский

---

Multi-step-ahead wind speed forecasting based on a hybrid decomposition method and temporal convolutional networks

Energy, Journal Year: 2021, Volume and Issue: 238, P. 121981 - 121981

Published: Sept. 7, 2021

Language: Английский

---

A review on the integrated optimization techniques and machine learning approaches for modeling, prediction, and decision making on integrated energy systems

Renewable Energy, Journal Year: 2022, Volume and Issue: 194, P. 822 - 849

Published: June 3, 2022

Language: Английский

---

Microgrid Digital Twins: Concepts, Applications, and Future Trends

IEEE Access, Journal Year: 2021, Volume and Issue: 10, P. 2284 - 2302

Published: Dec. 27, 2021

Following the fourth industrial revolution, and with recent advances in information communication technologies, digital twinning concept is attracting attention of both academia industry worldwide. A microgrid digital twin (MGDT) refers to representation a (MG), which mirrors behavior its physical counterpart by using high-fidelity models simulation platforms as well real-time bi-directional data exchange real twin. With massive deployment sensor networks IoT technologies MGs, huge volume continuously generated, contains valuable enhance performance MGs. MGDTs provide powerful tool manage historical stream an efficient secure manner support MGs’ operation assisting their design, management, maintenance. In this paper, (DT) key characteristics are introduced. Moreover, workflow for establishing presented. The goal explore different applications DTs namely control, operator training, forecasting, fault diagnosis, expansion planning, policy-making. Besides, up-to-date overview studies that applied DT power systems specifically MGs provided. Considering significance situational awareness, security, resilient potential enhancement light twinning thoroughly analyzed conceptual model management Finally, future trends discussed.

Language: Английский

---

Wind, Solar, and Photovoltaic Renewable Energy Systems with and without Energy Storage Optimization: A Survey of Advanced Machine Learning and Deep Learning Techniques

Energies, Journal Year: 2022, Volume and Issue: 15(2), P. 578 - 578

Published: Jan. 14, 2022

Nowadays, learning-based modeling methods are utilized to build a precise forecast model for renewable power sources. Computational Intelligence (CI) techniques have been recognized as effective in generating and optimizing tools. The complexity of this variety energy depends on its coverage large sizes data parameters, which be investigated thoroughly. This paper covered the most resent important researchers domain problems using methods. Various types Deep Learning (DL) Machine (ML) algorithms employed Solar Wind supplies given. performance given literature is assessed by new taxonomy. focus conducting comprehensive state-of-the-art heading evaluation discusses vital difficulties possibilities extensive research. Based results, variations efficiency, robustness, accuracy values, generalization capability obvious learning techniques. In case big dataset, effectiveness significantly better than other computational However, applying producing hybrid with optimization develop optimize construction optionally indicated. all cases, achievement single method due fact that gain benefit two or more providing an accurate forecast. Therefore, it suggested utilize future deal generation problems.

Language: Английский

---

Spatio-temporal wind speed forecasting using graph networks and novel Transformer architectures

Applied Energy, Journal Year: 2022, Volume and Issue: 333, P. 120565 - 120565

Published: Dec. 28, 2022

To improve the security and reliability of wind energy production, short-term forecasting has become utmost importance. This study focuses on multi-step spatio-temporal speed for Norwegian continental shelf. In particular, considers 14 offshore measurement stations aims to leverage spatial dependencies through relative physical location different local forecasts simultaneously output each locations. Our models produce either 10-minute, 1- or 4-hour forecasts, with 10-minute resolution, meaning that more informative time series predicted future trends. A graph neural network (GNN) architecture was used extract dependencies, update functions learn temporal correlations. These were implemented using architectures. One such architecture, Transformer, increasingly popular sequence modelling in recent years. Various alterations have been proposed better facilitate forecasting, which this focused Informer, LogSparse Transformer Autoformer. is first Autoformer applied any these Informer formulated a setting forecasting. By comparing against Long Short-Term Memory (LSTM) Multi-Layer Perceptron (MLP) models, showed altered architectures as GNNs able outperform these. Furthermore, we propose Fast Fourier (FFTransformer), novel based signal decomposition consists two separate streams analyse trend periodic components separately. The FFTransformer found achieve superior results 1-hour ahead significantly outperforming all other forecasts. code implement are made publicly available at: https://github.com/LarsBentsen/FFTransformer.

Language: Английский

---

Metaheuristic-Based Hyperparameter Tuning for Recurrent Deep Learning: Application to the Prediction of Solar Energy Generation

Axioms, Journal Year: 2023, Volume and Issue: 12(3), P. 266 - 266

Published: March 4, 2023

As solar energy generation has become more and important for the economies of numerous countries in last couple decades, it is highly to build accurate models forecasting amount green that will be produced. Numerous recurrent deep learning approaches, mainly based on long short-term memory (LSTM), are proposed dealing with such problems, but most may differ from one test case another respect architecture hyperparameters. In current study, use an LSTM a bidirectional (BiLSTM) data collection that, besides time series values denoting generation, also comprises corresponding information about weather. The research additionally endows hyperparameter tuning by means enhanced version recently metaheuristic, reptile search algorithm (RSA). output tuned neural network compared ones several other state-of-the-art metaheuristic optimization approaches applied same task, using experimental setup, obtained results indicate approach as better alternative. Moreover, best model achieved R2 0.604, normalized MSE value 0.014, which yields improvement around 13% over traditional machine models.

Language: Английский

---

CNN-LSTM vs. LSTM-CNN to Predict Power Flow Direction: A Case Study of the High-Voltage Subnet of Northeast Germany

Sensors, Journal Year: 2023, Volume and Issue: 23(2), P. 901 - 901

Published: Jan. 12, 2023

The massive installation of renewable energy sources together with storage in the power grid can lead to fluctuating consumption when there is a bi-directional flow due surplus electricity generation. To ensure security and reliability grid, high-quality prediction required. However, predicting remains challenge ever-changing characteristics influence weather on overcome these challenges, we present two most popular hybrid deep learning (HDL) models based combination convolutional neural network (CNN) long-term memory (LSTM) predict investigated cluster. In our approach, CNN-LSTM LSTM-CNN were trained different datasets terms size included parameters. aim was see whether dataset additional data affect performance proposed model flow. result shows that both achieve small error under certain conditions. While parameters training time accuracy HDL model.

Language: Английский

---

Learning based short term wind speed forecasting models for smart grid applications: An extensive review and case study

Electric Power Systems Research, Journal Year: 2023, Volume and Issue: 222, P. 109502 - 109502

Published: June 1, 2023

Language: Английский

---

A novel approach based on integration of convolutional neural networks and echo state network for daily electricity demand prediction

Energy, Journal Year: 2023, Volume and Issue: 275, P. 127430 - 127430

Published: April 8, 2023

Predicting electricity demand data is considered an essential task in decisions taking, and establishing new infrastructure the power generation network. To deliver a high-quality prediction, this paper proposes hybrid combination technique, based on deep learning model of Convolutional Neural Networks Echo State Networks, named as CESN. Daily from four sites (Roderick, Rocklea, Hemmant Carpendale), located Southeast Queensland, Australia, have been used to develop proposed prediction model. The study also analyzes five other machine learning-based models (support vector regression, multilayer perceptron, extreme gradient boosting, neural network, Light Gradient Boosting) compare evaluate outcomes approach. results obtained experimental showed that able obtain highest performance compared existing developed for daily forecasting. Based statistical approaches utilized study, approach presents accuracy among models. algorithm excellent accurate forecasting method, which outperformed state art algorithms are currently problem.

Language: Английский

---