Scholarlink

<p>MPPT is critical for maximizing PV system efficiency amidst growing global demand for sustainable energy. Conventional MPPT methods, such as P&amp;O, struggle with rapid irradiance and temperature changes or partial shading, leading to inefficiencies. To address these challenges, this study proposes a novel MPPT strategy based on the SAC reinforcement learning algorithm integrated with transfer learning. The proposed method enhances the adaptability and efficiency of PV systems by enabling fast, stable, and accurate power tracking under a wide range of operating scenarios, including partial shading and temperature fluctuations. A comprehensive Simulink model is developed, in which the SAC agent dynamically adjusts the converter’s duty cycle. The simulation results show that the proposed controller outperforms P&amp;O and DDPG methods, achieving up to 98.2% tracking efficiency with significantly reduced power oscillations. This is the first study to combine SAC with transfer learning for MPPT, offering a robust, adaptive, and scalable solution for real-time PV energy optimization.</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

Optimizing Maximum Power Point Tracking with Soft Actor-Critic Reinforcement Learning and Transfer Learning Techniques

  • 期刊论文

  • 发表日期：

    2025年12月2日

查看全文
  • Alireza Rostamipour,
  • Amir Hossein Ghayeni,
  • Mehdi Abdi

摘要

MPPT is critical for maximizing PV system efficiency amidst growing global demand for sustainable energy. Conventional MPPT methods, such as P&O, struggle with rapid irradiance and temperature changes or partial shading, leading to inefficiencies. To address these challenges, this study proposes a novel MPPT strategy based on the SAC reinforcement learning algorithm integrated with transfer learning. The proposed method enhances the adaptability and efficiency of PV systems by enabling fast, stable, and accurate power tracking under a wide range of operating scenarios, including partial shading and temperature fluctuations. A comprehensive Simulink model is developed, in which the SAC agent dynamically adjusts the converter’s duty cycle. The simulation results show that the proposed controller outperforms P&O and DDPG methods, achieving up to 98.2% tracking efficiency with significantly reduced power oscillations. This is the first study to combine SAC with transfer learning for MPPT, offering a robust, adaptive, and scalable solution for real-time PV energy optimization.