Ensemble learning for short circuit fault location estimation in distribution networks

OKUMUŞ H., NUROĞLU F. M., McLoone S.

Engineering Applications of Artificial Intelligence, cilt.159, 2025 (SCI-Expanded) identifier identifier

  • Yayın Türü: Makale / Tam Makale
  • Cilt numarası: 159
  • Basım Tarihi: 2025
  • Doi Numarası: 10.1016/j.engappai.2025.111640
  • Dergi Adı: Engineering Applications of Artificial Intelligence
  • Derginin Tarandığı İndeksler: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Academic Search Premier, Aerospace Database, Applied Science & Technology Source, Communication Abstracts, Compendex, Computer & Applied Sciences, INSPEC, Metadex, Civil Engineering Abstracts
  • Anahtar Kelimeler: Distribution network, Ensemble models, Fault location, Multilayer perceptron, Random forest
  • Karadeniz Teknik Üniversitesi Adresli: Evet

Özet

One of the most challenging tasks in power system operation is finding the exact location of a short circuit fault especially in distribution networks with branched structure. In this article a novel neural network ensemble model-based methodology is presented for fault location determination, which combines fault type classification, fault section identification and fault location estimation, and uses only 3 phase V-I measurements from a single sending end monitoring point as inputs. Two ensemble modelling paradigms are considered, namely, Neural Network (Multilayer Perceptron) Ensembles (NNE) and Random Forests (RF). Several different ensemble learning based structures are created using the proposed models and evaluated for fault location estimation on the IEEE-34 feeder benchmark. The average and maximum prediction errors under different fault conditions are used as performance metrics. The results for both the RF and stacked ensemble methods demonstrate that combining predictions enhances overall performance. State-of-the-art performance is achieved with a confidence-weighted stacked NNE-RF ensemble model.