Evaluating Machine Learning and Deep Learning models for predicting Wind Turbine power output from environmental factors

PLoS ONE, Год журнала: 2025, Номер 20(1), С. e0317619 - e0317619

Опубликована: Янв. 23, 2025

This study presents a comprehensive comparative analysis of Machine Learning (ML) and Deep (DL) models for predicting Wind Turbine (WT) power output based on environmental variables such as temperature, humidity, wind speed, direction. Along with Artificial Neural Network (ANN), Long Short-Term Memory (LSTM), Recurrent (RNN), Convolutional (CNN), the following ML were looked at: Linear Regression (LR), Support Vector Regressor (SVR), Random Forest (RF), Extra Trees (ET), Adaptive Boosting (AdaBoost), Categorical (CatBoost), Extreme Gradient (XGBoost), Light (LightGBM). Using dataset 40,000 observations, assessed R-squared, Mean Absolute Error (MAE), Root Square (RMSE). ET achieved highest performance among models, an R-squared value 0.7231 RMSE 0.1512. Among DL ANN demonstrated best performance, achieving 0.7248 0.1516. The results show that especially ANN, did slightly better than models. means they are at modeling non-linear dependencies in multivariate data. Preprocessing techniques, including feature scaling parameter tuning, improved model by enhancing data consistency optimizing hyperparameters. When compared to previous benchmarks, both demonstrates significant predictive accuracy gains WT forecasting. study’s novelty lies directly comparing diverse range algorithms while highlighting potential advanced computational approaches renewable energy optimization.

Язык: Английский

Integrating Evolutionary Game-Theoretical Methods and Deep Reinforcement Learning for Adaptive Strategy Optimization in User-Side Electricity Markets: A Comprehensive Review

Mathematics, Год журнала: 2024, Номер 12(20), С. 3241 - 3241

Опубликована: Окт. 16, 2024

With the rapid development of smart grids, strategic behavior evolution in user-side electricity market transactions has become increasingly complex. To explore dynamic mechanisms this area, paper systematically reviews application evolutionary game theory markets, focusing on its unique advantages modeling multi-agent interactions and strategy optimization. While excels explaining formation long-term stable strategies, it faces limitations when dealing with real-time changes high-dimensional state spaces. Thus, further investigates integration deep reinforcement learning, particularly Q-learning network (DQN), theory, aiming to enhance adaptability applications. The introduction DQN enables participants perform adaptive optimization rapidly changing environments, thereby more effectively responding supply–demand fluctuations markets. Through simulations based a model, study reveals characteristics under different conditions, highlighting interaction patterns among complex environments. In summary, comprehensive review not only demonstrates broad applicability markets but also extends potential decision making through modern algorithms, providing new theoretical foundations practical insights for future policy formulation.

Язык: Английский

Integrating Sustainability and Resilience Objectives for Energy Decisions: A Systematic Review

Resources, Год журнала: 2025, Номер 14(6), С. 97 - 97

Опубликована: Июнь 5, 2025

There is a need for simultaneous attention to sustainability and resilience objectives while making energy decisions because of the address disruptions or shocks that can result from system-wide changes due transitioning existing threats system performance. Owing this emerging research area, systematic review used Scopus database central question: What are trends practices enhance integration decisions? The articles peer-reviewed, empirical in field written English. Articles did not explicitly systems (or any value chains) gray literature were excluded study. final screening records resulted selection 75 effectively addressed decision objective, context, implementation (D-OCI), classification scheme supports 18 specific questions identify integrating objectives. highlighted advantageous evaluation provide valuable insights formulating policies. This particularly relevant energy-related affect households, organizations, both national international development. study proposes ideas future based on practices.

Язык: Английский