Scholarlink

<p>In recent years, renewable energy, particularly solar power, has attracted significant attention from both global and national authorities. Photovoltaic (PV) systems are central to this transition due to their safety, sustainability, and the abundant availability of sunlight. Accurate forecasting of Global Horizontal Irradiance (GHI) is critical for the performance and reliability of these systems. While hourly GHI prediction has improved, achieving precise minutely multi-step forecasts remains challenging. This study focuses on two Indian regions, Jaipur and Ajmer, and proposes a hybrid approach that combines the strengths of Federated Transformer and Informer models, further optimized using Grey Wolf Optimization (GWO). The proposed framework captures both short-term fluctuations and long-term patterns in irradiance data, ensuring higher accuracy and model stability. The experimental results clearly highlight the superior performance of the proposed model over benchmark architectures such as Auto former and Reformer. It achieves an impressive 26.0% improvement in <i>R</i><sup>2</sup> (0.96 for Jaipur and 0.92 for Ajmer) and an 11.2% reduction in Mean Absolute Error (4.88 W/m<sup>2</sup> for Jaipur and 7.12 W/m<sup>2</sup> for Ajmer). These outcomes underscore the model’s remarkable forecasting precision, demonstrating its strong potential for high-resolution solar prediction and significant enhancement in PV system efficiency and reliability.</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

Advancing Global Horizontal Irradiance Prediction with Grey Wolf Optimized Hybrid Transformer Models

  • 期刊论文

  • 发表日期：

    2025年11月17日

查看全文
  • Girijapati Sharma,
  • Rahul Gupta,
  • Rajendra Prasad Mahapatra

摘要

In recent years, renewable energy, particularly solar power, has attracted significant attention from both global and national authorities. Photovoltaic (PV) systems are central to this transition due to their safety, sustainability, and the abundant availability of sunlight. Accurate forecasting of Global Horizontal Irradiance (GHI) is critical for the performance and reliability of these systems. While hourly GHI prediction has improved, achieving precise minutely multi-step forecasts remains challenging. This study focuses on two Indian regions, Jaipur and Ajmer, and proposes a hybrid approach that combines the strengths of Federated Transformer and Informer models, further optimized using Grey Wolf Optimization (GWO). The proposed framework captures both short-term fluctuations and long-term patterns in irradiance data, ensuring higher accuracy and model stability. The experimental results clearly highlight the superior performance of the proposed model over benchmark architectures such as Auto former and Reformer. It achieves an impressive 26.0% improvement in R2 (0.96 for Jaipur and 0.92 for Ajmer) and an 11.2% reduction in Mean Absolute Error (4.88 W/m2 for Jaipur and 7.12 W/m2 for Ajmer). These outcomes underscore the model’s remarkable forecasting precision, demonstrating its strong potential for high-resolution solar prediction and significant enhancement in PV system efficiency and reliability.