Scholarlink

<p>In the digital era, Digital Twins (DTs) have emerged as a crucial tool for real-time optimization of photovoltaic (PV) system performance. Partial shading remains a major challenge, significantly reducing energy output in PV systems. This paper presents a novel methodology to enhance per-day energy extraction (PDEE) from PV systems under partial shading conditions. The approach involves two key steps: (1) Developing a Digital Twin framework for the PV system using real-world sensor data (voltage, current, temperature, and irradiance) for model training and calibration, enabling accurate online power estimation while reducing the reliance on physical sensors. (2) Integrating auxiliary (biased) sources in series with each partially shaded PV array to regulate the Global Maximum Power Point Tracking (GMPPT) voltage across all parallel-connected PV arrays, ensuring optimal power extraction. A digital twin of the single-diode PV model is constructed using a precise mathematical representation, with parameters estimated via deep neural networks (DNNs) optimized by the Modified Harris Hawks Optimization (MHHO) algorithm. While essential measurements such as temperature and irradiance are retained for real-time estimation, validation, and fault detection, the Digital Twin serves as an intelligent virtual replica of the physical PV system, significantly reducing reliance on extensive sensor networks. Additionally, a machine learning model, trained on features derived from the digital twin, is employed to predict the voltage at GMPPT accurately. Experimental validations and simulations demonstrate that the proposed methodology enhances power extraction by 15.86% under partial shading conditions. This approach offers a practical and effective solution for improving energy harvesting efficiency and mitigating the impact of partial shading in real-world PV applications.</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

Data-Driven Digital Twin for Intelligent Energy Optimization in Partially Shaded PV Systems

  • 期刊论文

  • 发表日期：

    2025年7月16日

查看全文
  • A. Aswini,
  • P. Sivakumar,
  • M. Kaleeswari

摘要

In the digital era, Digital Twins (DTs) have emerged as a crucial tool for real-time optimization of photovoltaic (PV) system performance. Partial shading remains a major challenge, significantly reducing energy output in PV systems. This paper presents a novel methodology to enhance per-day energy extraction (PDEE) from PV systems under partial shading conditions. The approach involves two key steps: (1) Developing a Digital Twin framework for the PV system using real-world sensor data (voltage, current, temperature, and irradiance) for model training and calibration, enabling accurate online power estimation while reducing the reliance on physical sensors. (2) Integrating auxiliary (biased) sources in series with each partially shaded PV array to regulate the Global Maximum Power Point Tracking (GMPPT) voltage across all parallel-connected PV arrays, ensuring optimal power extraction. A digital twin of the single-diode PV model is constructed using a precise mathematical representation, with parameters estimated via deep neural networks (DNNs) optimized by the Modified Harris Hawks Optimization (MHHO) algorithm. While essential measurements such as temperature and irradiance are retained for real-time estimation, validation, and fault detection, the Digital Twin serves as an intelligent virtual replica of the physical PV system, significantly reducing reliance on extensive sensor networks. Additionally, a machine learning model, trained on features derived from the digital twin, is employed to predict the voltage at GMPPT accurately. Experimental validations and simulations demonstrate that the proposed methodology enhances power extraction by 15.86% under partial shading conditions. This approach offers a practical and effective solution for improving energy harvesting efficiency and mitigating the impact of partial shading in real-world PV applications.