Prediction of solar energy guided by pearson correlation using machine learning. (1st June 2021)
- Record Type:
- Journal Article
- Title:
- Prediction of solar energy guided by pearson correlation using machine learning. (1st June 2021)
- Main Title:
- Prediction of solar energy guided by pearson correlation using machine learning
- Authors:
- Jebli, Imane
Belouadha, Fatima-Zahra
Kabbaj, Mohammed Issam
Tilioua, Amine - Abstract:
- Abstract: Solar energy forecasting represents a key element in increasing the competitiveness of solar power plants in the energy market and reducing the dependence on fossil fuels in economic and social development. This paper presents an approach for predicting solar energy, based on machine and deep learning techniques. The relevance of the studied models was evaluated for real-time and short-term solar energy forecasting to ensure optimized management and security requirements in this field while using an integral solution based on a single tool and an appropriate predictive model. The datasets we used in this study, represent data from 2016 to 2018 and are related to Errachidia which is a semi-desert climate province in Morocco. Pearson correlation coefficient was deployed to identify the most relevant meteorological inputs from which the models should learn. RF and ANN have provided high accuracies against LR and SVR, which have reported very significant errors. ANN has shown good performance for both real-time and short-term predictions. The key findings were compared with Pirapora in Brazil, which is a tropical climate region, to show the quality and reproducibility of the study. Graphical abstract: Image 1 Highlights: The efficiency of Pearson correlation coefficient was investigated. The accuracy of LR, RF, SVR and ANN for forecasting solar energy was evaluated. The performance was evaluated by MAE, MSE, RMSE, Max Error, R-squared and NRMSE. RF and ANN models showAbstract: Solar energy forecasting represents a key element in increasing the competitiveness of solar power plants in the energy market and reducing the dependence on fossil fuels in economic and social development. This paper presents an approach for predicting solar energy, based on machine and deep learning techniques. The relevance of the studied models was evaluated for real-time and short-term solar energy forecasting to ensure optimized management and security requirements in this field while using an integral solution based on a single tool and an appropriate predictive model. The datasets we used in this study, represent data from 2016 to 2018 and are related to Errachidia which is a semi-desert climate province in Morocco. Pearson correlation coefficient was deployed to identify the most relevant meteorological inputs from which the models should learn. RF and ANN have provided high accuracies against LR and SVR, which have reported very significant errors. ANN has shown good performance for both real-time and short-term predictions. The key findings were compared with Pirapora in Brazil, which is a tropical climate region, to show the quality and reproducibility of the study. Graphical abstract: Image 1 Highlights: The efficiency of Pearson correlation coefficient was investigated. The accuracy of LR, RF, SVR and ANN for forecasting solar energy was evaluated. The performance was evaluated by MAE, MSE, RMSE, Max Error, R-squared and NRMSE. RF and ANN models show high level of accuracy for real-time solar energy prediction. ANN outperforms all of the other models for short-term solar energy prediction. … (more)
- Is Part Of:
- Energy. Volume 224(2021)
- Journal:
- Energy
- Issue:
- Volume 224(2021)
- Issue Display:
- Volume 224, Issue 2021 (2021)
- Year:
- 2021
- Volume:
- 224
- Issue:
- 2021
- Issue Sort Value:
- 2021-0224-2021-0000
- Page Start:
- Page End:
- Publication Date:
- 2021-06-01
- Subjects:
- Solar energy prediction -- Machine and deep learning -- Linear regression -- Random forest -- Support vector regression -- Artificial neural networks
Power resources -- Periodicals
Power (Mechanics) -- Periodicals
Energy consumption -- Periodicals
333.7905 - Journal URLs:
- http://www.elsevier.com/journals ↗
- DOI:
- 10.1016/j.energy.2021.120109 ↗
- Languages:
- English
- ISSNs:
- 0360-5442
- Deposit Type:
- Legaldeposit
- View Content:
- Available online (eLD content is only available in our Reading Rooms) ↗
- Physical Locations:
- British Library DSC - 3747.445000
British Library DSC - BLDSS-3PM
British Library HMNTS - ELD Digital store - Ingest File:
- 25582.xml