Probabilistic solar irradiance forecasting based on XGBoost

Energy Reports, Journal Year: 2022, Volume and Issue: 8, P. 1087 - 1095

Published: March 14, 2022

Solar energy has received increasing attention as renewable clean in recent years. Power grid operators and researchers widely value probabilistic solar irradiance forecasting because it can provide uncertainty measurement for future PV production. This paper proposes a prediction model of based on XGBoost. Specifically, after data preprocessing, historical is utilized training point Since XGBoost obtained by minimizing the residuals successive iterations multiple trees, when predicting at certain time future, these trees generate predicted values iteratively. Finally, kernel density estimation method applied to transform above results probability intervals under different confidence levels. Experimental public sets show that this better accuracy than other benchmark algorithms. The experiment also shows proposed requires less simple parameter adjustment, which very suitable application engineering practice.

Language: Английский

---

Solar photovoltaic power prediction using artificial neural network and multiple regression considering ambient and operating conditions

Energy Conversion and Management, Journal Year: 2023, Volume and Issue: 288, P. 117186 - 117186

Published: May 18, 2023

Language: Английский

---

Modelling and real time performance evaluation of a 5 MW grid-connected solar photovoltaic plant using different artificial neural networks

Energy Conversion and Management, Journal Year: 2023, Volume and Issue: 279, P. 116767 - 116767

Published: Feb. 10, 2023

Language: Английский

---

A novel learning approach for short-term photovoltaic power forecasting - A review and case studies

Engineering Applications of Artificial Intelligence, Journal Year: 2024, Volume and Issue: 133, P. 108502 - 108502

Published: April 29, 2024

Language: Английский

---

Advancing solar PV panel power prediction: A comparative machine learning approach in fluctuating environmental conditions

Case Studies in Thermal Engineering, Journal Year: 2024, Volume and Issue: 59, P. 104459 - 104459

Published: May 1, 2024

Solar photovoltaic (PV) panels play a crucial role in sustainable energy generation, yet their power output often faces uncertainties due to dynamic weather conditions. In this study, comparative machine learning approach is introduced, utilizing multivariate regression (MR), support vector (SVMR), and Gaussian (GR) techniques for precise solar PV panel prediction. The investigation into the impact of environmental factors—solar radiation, ambient temperature, relative humidity—on reveals superior predictive capabilities SVMR models. With mean squared error (MSE) 0.038, absolute (MAE) 0.17, an R2 value 0.99, outperforms GR MR Conversely, demonstrates comparatively weaker performance, yielding 0.88, MSE 0.49, MAE 0.63. This research underscores reliability enhanced accuracy proposed model forecasting output. outcomes presented herein carry significant implications promoting widespread adoption electricity particularly challenging findings offer valuable insights optimizing deployment, ultimately contributing expansion generation national landscape. Moreover, analysis provides how anticipated can adapt varying conditions, encompassing factors such as humidity, radiation.

Language: Английский

---

Forecasting Solar Photovoltaic Power Production: A Comprehensive Review and Innovative Data-Driven Modeling Framework

Energies, Journal Year: 2024, Volume and Issue: 17(16), P. 4145 - 4145

Published: Aug. 20, 2024

The intermittent and stochastic nature of Renewable Energy Sources (RESs) necessitates accurate power production prediction for effective scheduling grid management. This paper presents a comprehensive review conducted with reference to pioneering, comprehensive, data-driven framework proposed solar Photovoltaic (PV) generation prediction. systematic integrating comprises three main phases carried out by seven modules addressing numerous practical difficulties the task: phase I handles aspects related data acquisition (module 1) manipulation 2) in preparation development scheme; II tackles associated model 3) assessment its accuracy 4), including quantification uncertainty 5); III evolves towards enhancing incorporating context change detection 6) incremental learning when new become available 7). adeptly addresses all facets PV prediction, bridging existing gaps offering solution inherent challenges. By seamlessly these elements, our approach stands as robust versatile tool precision real-world applications.

Language: Английский

---

Ensemble Approach of Optimized Artificial Neural Networks for Solar Photovoltaic Power Prediction

IEEE Access, Journal Year: 2019, Volume and Issue: 7, P. 81741 - 81758

Published: Jan. 1, 2019

The use of data-driven ensemble approaches for the prediction solar Photovoltaic (PV) power production is promising due to their capability handling intermittent nature energy source. In this work, a comprehensive approach composed by optimized and diversified Artificial Neural Networks (ANNs) proposed improving 24h-ahead PV predictions. ANNs are in terms number hidden neurons diverse training datasets used build ANNs, resorting trial-and-error procedure BAGGING techniques, respectively. addition, Bootstrap technique embedded quantifying sources uncertainty that affect models' predictions form Prediction Intervals (PIs). effectiveness demonstrated real case study regarding grid-connected system (231 kWac capacity) installed on rooftop Faculty Engineering at Applied Science Private University (ASU), Amman, Jordan. results show outperforms three benchmark models, including smart persistence model single ANN currently adopted system's owner task, with performance gain reaches up 11%, 12%, 9%, RMSE, MAE, WMAE standard metrics, Simultaneously, has shown superior affecting predictions, establishing slightly wider PIs achieve highest confidence level 84% predefined 80% compared other literature. These enhancements would, indeed, allow balancing supplies demands across centralized grid networks through economic dispatch decisions between contribute mix.

Language: Английский

---

Big Data for Energy Management and Energy-Efficient Buildings

Energies, Journal Year: 2020, Volume and Issue: 13(7), P. 1555 - 1555

Published: March 27, 2020

European buildings are producing a massive amount of data from wide spectrum energy-related sources, such as smart meters’ data, sensors and other Internet things devices, creating new research challenges. In this context, the aim paper is to present high-level data-driven architecture for exchange, management real-time processing. This multi-disciplinary big environment enables integration cross-domain combined with emerging artificial intelligence algorithms distributed ledgers technology. Semantically enhanced, interlinked multilingual repositories heterogeneous types coupled set visualization, querying exploration tools, suitable application programming interfaces (APIs) well suite configurable ready-to-use analytical components that implement series advanced machine learning deep algorithms. The results pilot proposed framework presented discussed. reliable effective policymaking, supports creation exploitation innovative energy efficiency services through utilization variety operation buildings.

Language: Английский

---

A minutely solar irradiance forecasting method based on real-time sky image-irradiance mapping model

Energy Conversion and Management, Journal Year: 2020, Volume and Issue: 220, P. 113075 - 113075

Published: June 23, 2020

Language: Английский

---

Forecasting of Wastewater Treatment Plant Key Features Using Deep Learning-Based Models: A Case Study

IEEE Access, Journal Year: 2020, Volume and Issue: 8, P. 184475 - 184485

Published: Jan. 1, 2020

The accurate forecast of wastewater treatment plant (WWTP) key features can comprehend and predict the behavior to support process design controls, improve system reliability, reduce operational costs, endorse optimization overall performances. Deep learning technologies as proven data-driven soft-sensors should be developed for WWTP applications tackle non-linearity dynamic nature environmental data. This study adopts deep learning-based models features, such influent flow, temperature, biochemical oxygen demand (BOD), effluent chloride, BOD, power consumption. We constructed six derived from long short-term memory (LSTM) gated recurrent unit (GRU), namely traditional LSTM GRU, exponentially smoothed LSTM, adaptive version LSTM. employment a technique is expected outlier effect forecasting accuracy. Meanwhile, usage will enhance capabilities quickly accurately follow trend future compared performance these with Bi-directional (BiLSTM) seasonal decomposition using local regression. historical records coastal municipal in Saudi Arabia are used verify investigated models' effectiveness. proposed provide promising results but require no assumptions on data distributions. In terms efficiency, GRU based converge faster than models. accuracy, soft-sensor shows optimal result all followed by exponentially-smoothed By contrast, achieved lowest other These findings benefit practitioners achieve management.

Language: Английский

---