Scholarlink

<p>The global imperative for sustainable development increasingly pivots on the expansion and integration of solar energy. In this context, satellite remote sensing emerges as an indispensable tool, providing robust capabilities for monitoring diverse environmental features, notably cloud cover (CC). This paper proposes a deep-based intelligent framework for cloud detection and analysis, crucial for enhanced solar power generation. Based on the standard YOLOv5s model, the C3SE-YOLOv5s object detection model is improved by integrating the Squeeze-and-Excitation (SE) attention mechanism into the C3 module. The results demonstrate that the C3SE-YOLOv5s model outperforms the baseline YOLOv5s model, achieving a 3.4% increase in mAP, a 4.4% increase in precision, and a 5,6% increase in recall. Additionally, C3SE-YOLOv5s offers a 13% speed improvement over YOLOv5s and a 72% speed improvement over YOLOv5x. The deep intelligent framework in this study incorporates image processing techniques, provides precise localization of expanding clouds, assesses shading risks, and analyzes cloud formation trends. The results significantly advance the field of renewable energy by enhancing solar power system efficiency, improving the understanding of environmental impacts on photovoltaic technology.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing photovoltaic plant efficiency through a deep framework for cloud detection, shading risk assessment, and cloud trend analysis

  • 期刊论文

  • 发表日期：

    2025年11月3日

查看全文
  • Mohamed Khala,
  • Omar Eloutassi

摘要

The global imperative for sustainable development increasingly pivots on the expansion and integration of solar energy. In this context, satellite remote sensing emerges as an indispensable tool, providing robust capabilities for monitoring diverse environmental features, notably cloud cover (CC). This paper proposes a deep-based intelligent framework for cloud detection and analysis, crucial for enhanced solar power generation. Based on the standard YOLOv5s model, the C3SE-YOLOv5s object detection model is improved by integrating the Squeeze-and-Excitation (SE) attention mechanism into the C3 module. The results demonstrate that the C3SE-YOLOv5s model outperforms the baseline YOLOv5s model, achieving a 3.4% increase in mAP, a 4.4% increase in precision, and a 5,6% increase in recall. Additionally, C3SE-YOLOv5s offers a 13% speed improvement over YOLOv5s and a 72% speed improvement over YOLOv5x. The deep intelligent framework in this study incorporates image processing techniques, provides precise localization of expanding clouds, assesses shading risks, and analyzes cloud formation trends. The results significantly advance the field of renewable energy by enhancing solar power system efficiency, improving the understanding of environmental impacts on photovoltaic technology.