Scholarlink

Accurate short-term forecasting of power output of Floating Photovoltaic (FPV) systems play a critical role in enhancing renewable energy generation efficiency and refining energy management strategies. This paper investigates the performance of a well-known type of recurrent neural network, the Long Short-Term Memory, applied to the short-term forecast of the power generation of two FPV power plants: a horizontal-axis tracking bifacial FPV system and a East-West bifacial FPV system, both installed at the “Enel InnovationHub&Lab” in Catania, Italy. To explain the behavior of the developed forecasting model and identify the variables most strongly correlated with the model’s output, a feature importance analysis has been conducted using SHAP (SHapley Additive exPlanations). The results highlight the effectiveness of the developed methodology in forecasting FPV power output, supporting the integration of FPV systems into the grid by enhancing the operational performance and grid stability.

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

Short-Term AC Power Forecasting for Floating Bifacial Photovoltaic Systems Using LSTM Networks

  • 会议论文

  • 发表日期：

    2026年

查看全文
  • Cristina Ventura,
  • Giuseppe Marco Tina,
  • Giovanni Bellitto,
  • Concetto Spampinato,
  • Sergio Ferlito,
  • Girolamo Di Francia

摘要

Accurate short-term forecasting of power output of Floating Photovoltaic (FPV) systems play a critical role in enhancing renewable energy generation efficiency and refining energy management strategies. This paper investigates the performance of a well-known type of recurrent neural network, the Long Short-Term Memory, applied to the short-term forecast of the power generation of two FPV power plants: a horizontal-axis tracking bifacial FPV system and a East-West bifacial FPV system, both installed at the “Enel InnovationHub&Lab” in Catania, Italy. To explain the behavior of the developed forecasting model and identify the variables most strongly correlated with the model’s output, a feature importance analysis has been conducted using SHAP (SHapley Additive exPlanations). The results highlight the effectiveness of the developed methodology in forecasting FPV power output, supporting the integration of FPV systems into the grid by enhancing the operational performance and grid stability.