Two-Tier Cascaded Classifiers to Improve Electrical Power Quality


ELMASRY W., Wadi M., Shahinzadeh H.

26th International Electrical Power Distribution Conference, EPDC 2022, Tehran, Iran, 11 - 12 May 2022, pp.96-101 identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1109/epdc56235.2022.9817239
  • City: Tehran
  • Country: Iran
  • Page Numbers: pp.96-101
  • Keywords: Binary classifiers, Distribution networks, Electrical fault classification, Electrical fault detection, Electrical Power quality, Feature and hyperparameters optimization
  • Istanbul Kültür University Affiliated: Yes

Abstract

Although the manifestation of faults in electrical distribution networks is likely scarce, they deemed to be one of the most severe threats which encounter the reliability and stability of power systems. Thereby, most of research efforts in the last decade have been directed to characterize the faults and propose detection systems. However, the state-of-art of these systems still undergo serious shortcomings such as automation, integration and validation. To surpass these shortcomings, a two-tier system is proposed for both fault detection and fault classification in electrical distribution networks. Prior to deal with faults, the captured voltage signals pass through a preparatory unit to process the raw voltage signals. Afterwards, the optimized classifiers detect and classify the faults in the first and second tiers, respectively. Furthermore, the performance of the classifiers in both tiers is validated using the VSB dataset (real-time data for fault detection). Finally, the proposed two-tier system proved its efficiency in detecting electrical faults and their type as well.