Advancing solar PV panel power prediction: A comparative machine learning approach in fluctuating environmental conditions
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, a comparative machine learning approach is introduced, utilizing multivariate regression (MR), support vector machin...
Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Article |
| Language: | English English |
| Published: |
Elsevier Ltd
2024
|
| Subjects: | |
| Online Access: | http://irep.iium.edu.my/112174/2/112174_Advancing%20solar%20PV%20panel%20power%20prediction.pdf http://irep.iium.edu.my/112174/8/112174_Advancing%20solar%20PV%20panel%20power%20prediction_SCOPUS.pdf http://irep.iium.edu.my/112174/ https://www.sciencedirect.com/science/article/pii/S2214157X24004908/pdfft?md5=7abfe3686998480f7d3783fe557a5b8f&pid=1-s2.0-S2214157X24004908-main.pdf https://doi.org/10.1016/j.csite.2024.104459 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | 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, a comparative machine learning approach is introduced, utilizing multivariate regression (MR), support vector machine regression (SVMR), and Gaussian regression (GR) techniques for precise solar PV panel power prediction. The investigation into the impact of environmental factors—solar radiation, ambient temperature, and relative humidity—on PV panel output reveals the superior predictive capabilities of SVMR models. With a mean squared error (MSE) of 0.038, a mean absolute error (MAE) of 0.17, and an R-value of 0.99, SVMR outperforms GR and MR models. Conversely, Gaussian regression demonstrates comparatively weaker performance, yielding an R of 0.88, an MSE of 0.49, and an MAE of 0.63. This research underscores the reliability and enhanced accuracy of the proposed SVMR model in forecasting solar PV panel output. The outcomes presented herein carry significant implications for promoting the widespread adoption of PV panels in electricity generation, particularly in challenging environmental conditions. The findings offer valuable insights into optimizing solar PV deployment, ultimately contributing to the expansion of solar power generation in the national energy landscape. Moreover, the comparative analysis provides insights into how anticipated PV power generation can adapt to varying weather conditions, encompassing factors such as temperature, humidity, and solar radiation. |
|---|