Scholarlink

<p>The contemporary energy landscape, driven by the urgency to mitigate environmental pollution and the decline in fossil fuel reserves, has underscored the significance of renewable energy, particularly solar power, as a vital factor for future sustainable growth. This paper explores the challenges and techniques associated with forecasting photovoltaic (PV) power, with a focus on the short term. Amidst the current prediction models—the physical model, time series model, machine learning model, and the hybrid method—this study introduces an innovative machine learning approach for short-term PV power forecasting, combining the prowess of AdaBoost and Decision Trees (DT). A unique aspect of this research is the use of the MRFO optimizer to refine the hybrid model further. In this framework, MRFO (Manta Ray Foraging Optimization) is employed to automatically optimize key hyperparameters, enabling the hybrid models to achieve higher predictive accuracy and better generalization across different sites. Data derived from three distinct U.S. sites—encompassing a comprehensive set of meteorological variables—formed the foundation for model training and evaluation. The results consistently revealed the AdaBoost-MRFO hybrid model’s superior accuracy, achieving R<sup>2</sup> values of 0.981 for March AFB, 0.980 for MNANG, and 0.977 for Offutt, thereby outperforming the DT-MRFO model in all three cases. This study not only enriches the current forecasting methodologies but also highlights the machine learning approach’s capability in revolutionizing the renewable energy sector.</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

Accurate Short-Term Photovoltaic (PV) Power Forecasting: Leveraging the Capabilities of AdaBoost and Decision Trees

  • 期刊论文

  • 发表日期：

    2026年4月28日

查看全文
  • Xingbo Liu,
  • Jiajia Zhan,
  • Liu Xinyue

摘要

The contemporary energy landscape, driven by the urgency to mitigate environmental pollution and the decline in fossil fuel reserves, has underscored the significance of renewable energy, particularly solar power, as a vital factor for future sustainable growth. This paper explores the challenges and techniques associated with forecasting photovoltaic (PV) power, with a focus on the short term. Amidst the current prediction models—the physical model, time series model, machine learning model, and the hybrid method—this study introduces an innovative machine learning approach for short-term PV power forecasting, combining the prowess of AdaBoost and Decision Trees (DT). A unique aspect of this research is the use of the MRFO optimizer to refine the hybrid model further. In this framework, MRFO (Manta Ray Foraging Optimization) is employed to automatically optimize key hyperparameters, enabling the hybrid models to achieve higher predictive accuracy and better generalization across different sites. Data derived from three distinct U.S. sites—encompassing a comprehensive set of meteorological variables—formed the foundation for model training and evaluation. The results consistently revealed the AdaBoost-MRFO hybrid model’s superior accuracy, achieving R2 values of 0.981 for March AFB, 0.980 for MNANG, and 0.977 for Offutt, thereby outperforming the DT-MRFO model in all three cases. This study not only enriches the current forecasting methodologies but also highlights the machine learning approach’s capability in revolutionizing the renewable energy sector.