Scholarlink

<p>Artificial intelligence (AI) is revolutionizing energy systems by addressing the challenges associated with integrating renewable energy sources (RES) into smart grids. As conventional energy sources decline, the transition to RES such as solar and wind power is vital for sustainable energy solutions. However, accurately predicting energy production and improving grid performance remain significant technological challenges. This study leverages real-time energy production data (measured in MWh) from solar PV, solar thermal, and wind plants over one year, incorporating features such as plane of array, heat, and wind speed. Predictions are initially generated using six advanced AI models like random forest, gradient boosting machines, autoregressive conditional heteroskedasticity, temporal convolutional networks, bidirectional encoder representations, and echo state networks. A meta-learning approach is then applied to identify the most effective combinations of these models, resulting in the temporal fusion ensemble and sequential gradient ensemble, which achieve error rates of 15.05% and 19.18%, respectively. To further enhance prediction accuracy, a parallel fusion technique is employed, reducing overall errors by 8.14% and demonstrating superior performance in energy forecasting. The results highlight the potential of hybrid AI-driven approaches to optimize energy resource management, improve grid reliability, and advance the state of predictive models for RES integration into smart grids, paving the way for more efficient and sustainable energy systems.</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

Multi-renewable energy resources parameters prediction through meta-learning models selectivity analysis and parallel fusion approaches

  • 期刊论文

  • 发表日期：

    2025年2月27日

查看全文
  • Muhammad Abubakar,
  • Yanbo Che,
  • Muhammad Shoaib Bhutta,
  • Ahsan Zafar

摘要

Artificial intelligence (AI) is revolutionizing energy systems by addressing the challenges associated with integrating renewable energy sources (RES) into smart grids. As conventional energy sources decline, the transition to RES such as solar and wind power is vital for sustainable energy solutions. However, accurately predicting energy production and improving grid performance remain significant technological challenges. This study leverages real-time energy production data (measured in MWh) from solar PV, solar thermal, and wind plants over one year, incorporating features such as plane of array, heat, and wind speed. Predictions are initially generated using six advanced AI models like random forest, gradient boosting machines, autoregressive conditional heteroskedasticity, temporal convolutional networks, bidirectional encoder representations, and echo state networks. A meta-learning approach is then applied to identify the most effective combinations of these models, resulting in the temporal fusion ensemble and sequential gradient ensemble, which achieve error rates of 15.05% and 19.18%, respectively. To further enhance prediction accuracy, a parallel fusion technique is employed, reducing overall errors by 8.14% and demonstrating superior performance in energy forecasting. The results highlight the potential of hybrid AI-driven approaches to optimize energy resource management, improve grid reliability, and advance the state of predictive models for RES integration into smart grids, paving the way for more efficient and sustainable energy systems.