Scholarlink

<p>This research presented an automated model for fault detection of grid grid-connected solar photovoltaic (PV) systems with an improvement in fault detection in grid-connected solar power systems with the <i>k</i>-nearest neighbors (KNN) algorithm using feature selection. The dataset consists of data in two modes, i.e., low power point tracking (LPPT) and maximum power point tracking (MPPT) for each fault. Seven faults are considered in the research which includes a vast range of complexity which is a challenge for the practical grid-connected PV systems. Using feature selection, the number of independent variables is reduced as well as the performance of the algorithm is enhanced to 0.999995 in terms of <i>F</i>1 score. Profiles for parameters for each fault in the solar PV system have been created. A comparison of the KNN algorithm is also done with the artificial neural network (ANN) algorithm on the same dataset to verify the performance. The method outperforms principal component analysis (PCA) for dimensionality reduction which is a computationally heavy process. Therefore, the research overall improves the performance of solar systems by automatically fault detection and it reduces complexity and reduces cost too in solar power generation.</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

Automated model for fault detection in grid-connected solar systems

  • 期刊论文

  • 出版模式：

    开放获取

  • 发表日期：

    2025年2月27日

查看全文
  • Vinay Singh,
  • Ruby Beniwal

摘要

This research presented an automated model for fault detection of grid grid-connected solar photovoltaic (PV) systems with an improvement in fault detection in grid-connected solar power systems with the k-nearest neighbors (KNN) algorithm using feature selection. The dataset consists of data in two modes, i.e., low power point tracking (LPPT) and maximum power point tracking (MPPT) for each fault. Seven faults are considered in the research which includes a vast range of complexity which is a challenge for the practical grid-connected PV systems. Using feature selection, the number of independent variables is reduced as well as the performance of the algorithm is enhanced to 0.999995 in terms of F1 score. Profiles for parameters for each fault in the solar PV system have been created. A comparison of the KNN algorithm is also done with the artificial neural network (ANN) algorithm on the same dataset to verify the performance. The method outperforms principal component analysis (PCA) for dimensionality reduction which is a computationally heavy process. Therefore, the research overall improves the performance of solar systems by automatically fault detection and it reduces complexity and reduces cost too in solar power generation.