Short-Medium-Term Solar Irradiance Forecasting with a CEEMDAN-CNN-ATT-LSTM Hybrid Model Using Meteorological Data

Applied Sciences, Год журнала: 2025, Номер 15(3), С. 1275 - 1275

Опубликована: Янв. 26, 2025

In recent years, the adverse effects of climate change have increased rapidly worldwide, driving countries to transition clean energy sources such as solar and wind. However, these energies face challenges cloud cover, precipitation, wind speed, temperature, which introduce variability intermittency in power generation, making integration into interconnected grid difficult. To achieve this, we present a novel hybrid deep learning model, CEEMDAN-CNN-ATT-LSTM, for short- medium-term irradiance prediction. The model utilizes complete empirical ensemble modal decomposition with adaptive noise (CEEMDAN) extract intrinsic seasonal patterns irradiance. addition, it employs encoder-decoder framework that combines convolutional neural networks (CNN) capture spatial relationships between variables, an attention mechanism (ATT) identify long-term patterns, long short-term memory (LSTM) network dependencies time series data. This has been validated using meteorological data more than 2400 masl region characterized by complex climatic conditions south Ecuador. It was able predict at 1, 6, 12 h horizons, mean absolute error (MAE) 99.89 W/m2 winter 110.13 summer, outperforming reference methods this study. These results demonstrate our represents progress contributing scientific community field environments high its applicability real scenarios.

Язык: Английский

---

Evaluation of Solar Radiation Prediction Models Using AI: A Performance Comparison in the High-Potential Region of Konya, Türkiye

Atmosphere, Год журнала: 2025, Номер 16(4), С. 398 - 398

Опубликована: Март 30, 2025

Solar radiation is one of the most abundant energy sources in world and a crucial parameter that must be researched developed for sustainable projects future generations. This study evaluates performance different machine learning methods solar prediction Konya, Turkey, region with high potential. The analysis based on hydro-meteorological data collected from NASA/POWER, covering period 1 January 1984 to 31 December 2022. compares Long Short-Term Memory (LSTM), Bidirectional LSTM (Bi-LSTM), Gated Recurrent Unit (GRU), GRU (Bi-GRU), LSBoost, XGBoost, Bagging, Random Forest (RF), General Regression Neural Network (GRNN), Support Vector Machines (SVM), Artificial Networks (MLANN, RBANN). variables used include temperature, relative humidity, precipitation, wind speed, while target variable radiation. dataset was divided into 75% training 25% testing. Performance evaluations were conducted using Mean Absolute Error (MAE), Root Square (RMSE), coefficient determination (R2). results indicate Bi-LSTM models performed best test phase, demonstrating superiority deep learning-based approaches prediction.

Язык: Английский

---

A multistep short-term solar radiation forecasting model using fully convolutional neural networks and chaotic aquila optimization combining WRF-Solar model results

Energy, Год журнала: 2023, Номер 271, С. 126980 - 126980

Опубликована: Фев. 20, 2023

Язык: Английский

---

Solar Irradiation Forecasting Using Ensemble Voting Based on Machine Learning Algorithms

Sustainability, Год журнала: 2023, Номер 15(10), С. 7943 - 7943

Опубликована: Май 12, 2023

This paper proposes an ensemble voting model for solar radiation forecasting based on machine learning algorithms. Several models are assessed using a simple average and weighted average, combining the following algorithms: random forest, extreme gradient boosting, categorical adaptive boosting. A clustering algorithm is used to group data according weather, feature selection applied choose most-related inputs their past observation values. Prediction performance evaluated by several metrics real-world Brazilian database, considering different prediction time horizons of up 12 h ahead. Numerical results show approach forest boosting has superior performance, with reduction 6% MAE, 3% RMSE, 16% MAPE, 1% R2 when predicting one hour in advance, outperforming individual algorithms other models.

Язык: Английский

---

Hourly solar radiation estimation and uncertainty quantification using hybrid models

Renewable and Sustainable Energy Reviews, Год журнала: 2024, Номер 202, С. 114727 - 114727

Опубликована: Июль 4, 2024

Язык: Английский

---

A new evolutionary forest model via incremental tree selection for short-term global solar irradiance forecasting under six various climatic zones

Energy Conversion and Management, Год журнала: 2024, Номер 310, С. 118471 - 118471

Опубликована: Апрель 30, 2024

Язык: Английский

---

Deep learning performance prediction for solar-thermal-driven hydrogen production membrane reactor via bayesian optimized LSTM

International Journal of Hydrogen Energy, Год журнала: 2024, Номер 82, С. 1402 - 1412

Опубликована: Авг. 9, 2024

Язык: Английский

---

Prospects and Challenges of Energy Storage Materials: A Comprehensive Review

Chemical Engineering Journal Advances, Год журнала: 2024, Номер 20, С. 100657 - 100657

Опубликована: Окт. 10, 2024

Язык: Английский

---

On the use of sky images for intra-hour solar forecasting benchmarking: Comparison of indirect and direct approaches

Solar Energy, Год журнала: 2024, Номер 276, С. 112649 - 112649

Опубликована: Июнь 6, 2024

Язык: Английский

---

Enhancing Solar Forecasting Accuracy with Sequential Deep Artificial Neural Network and Hybrid Random Forest and Gradient Boosting Models across Varied Terrains

Advanced Theory and Simulations, Год журнала: 2024, Номер 7(7)

Опубликована: Апрель 30, 2024

Abstract Effective solar energy utilization demands improvements in forecasting due to the unpredictable nature of irradiance (SI). This study introduces and rigorously tests two innovative models across different locations: Sequential Deep Artificial Neural Network (SDANN) Hybrid Random Forest Gradient Boosting (RFGB). SDANN, leveraging deep learning, aims identify complex patterns weather data, while RFGB, combining Boosting, proves more effective by offering a superior balance efficiency accuracy. The research highlights SDANN model's learning capabilities along with RFGB unique blend their comparative success over existing such as eXtreme (XGBOOST), Categorical (CatBOOST), Gated Recurrent Unit (GRU), K‐Nearest Neighbors (KNN) XGBOOST hybrid. With lowest Mean Squared Error (147.22), Absolute (8.77), high R 2 value (0.80) studied region, stands out. Additionally, detailed ablation studies on meteorological feature impacts model performance further enhance accuracy adaptability. By integrating cutting‐edge AI SI forecasting, this not only advances field but also sets stage for future renewable strategies global policy‐making.

Язык: Английский

---