Scholarlink

<p>The network topology plays a crucial role in ensuring the secure operation of a low-voltage (LV) distribution network, but its accuracy is limited by measurement precision. This paper proposes a method to enhance topology identification accuracy through feature selection and multi-feature decision-making. A multi-model fusion approach is introduced by combining several existing topology identification models to generate multiple results. Feature selection is applied to identify the simplest topology combination, and multi-feature decision-making resolves conflicts within this combination, ultimately providing a complete topology structure. The experimental results show that energy feature methods perform well at line branch side, while voltage feature methods are effective at metering side. The proposed method combines the strengths of both features, MFBD achieves an accuracy of 92.86%, surpassing MLRE (89.5%) and SBC methods (33%–42%), achieving an accuracy exceeding 92%, approximately 8% improvement over traditional single-feature methods, based on real-world data from an LV distribution network with approximately 50 users and 280 data points. Simulation experiments further validate that feature selection can be effectively applied to multi-feature topology identification, demonstrating its adaptability in dynamic network conditions. The main contributions of this paper: The proposal of feature selection principles and methods for multiple topology identification results; The development of a multi-feature decision-making method for topology identification opens up new perspectives in LV distribution network research and provides potential insights for future studies on multi-model and incremental topology identification.</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

Low-voltage distribution network topology identification based on feature selection and multi-feature decision-making

  • 期刊论文

  • 发表日期：

    2025年6月6日

查看全文
  • Jiahui Lv,
  • Nan Zhang,
  • Yueshu Xie,
  • Haifeng Liu

摘要

The network topology plays a crucial role in ensuring the secure operation of a low-voltage (LV) distribution network, but its accuracy is limited by measurement precision. This paper proposes a method to enhance topology identification accuracy through feature selection and multi-feature decision-making. A multi-model fusion approach is introduced by combining several existing topology identification models to generate multiple results. Feature selection is applied to identify the simplest topology combination, and multi-feature decision-making resolves conflicts within this combination, ultimately providing a complete topology structure. The experimental results show that energy feature methods perform well at line branch side, while voltage feature methods are effective at metering side. The proposed method combines the strengths of both features, MFBD achieves an accuracy of 92.86%, surpassing MLRE (89.5%) and SBC methods (33%–42%), achieving an accuracy exceeding 92%, approximately 8% improvement over traditional single-feature methods, based on real-world data from an LV distribution network with approximately 50 users and 280 data points. Simulation experiments further validate that feature selection can be effectively applied to multi-feature topology identification, demonstrating its adaptability in dynamic network conditions. The main contributions of this paper: The proposal of feature selection principles and methods for multiple topology identification results; The development of a multi-feature decision-making method for topology identification opens up new perspectives in LV distribution network research and provides potential insights for future studies on multi-model and incremental topology identification.