Advanced Fault Detection and Classification in Distribution Systems Using PMUs and Optimized SVM & KNN Algorithms
Guardado en:
| Autores: | , , , |
|---|---|
| Formato: | póster de congreso |
| Fecha de Publicación: | 2025 |
| Descripción: | This paper presents an advanced methodology focused on fault detection and classification in distribution systemsusing real-time phasor measurements from virtual µPMUs andoptimized machine learning algorithms. A modified IEEE 24-busfeeder is simulated in OPAL-RT’s HYPERSIM, incorporatingdirectional overcurrent protection (50/51 and 67). SVM and KNNmodels are enhanced through feature selection and hyperparameter tuning offline via scikit-learn, with dimensionality reductionusing PCA. Results demonstrate significant improvements inclassification accuracy and speed for single-, double-, and threephase faults, validating the use of high-resolution synchrophasordata to support advanced protection in modern distributionnetworks. |
| País: | Kérwá |
| Institución: | Universidad de Costa Rica |
| Repositorio: | Kérwá |
| Lenguaje: | Inglés |
| OAI Identifier: | oai:kerwa.ucr.ac.cr:10669/104924 |
| Acceso en línea: | https://hdl.handle.net/10669/104924 https://doi.org/10.1109/CONCAPAN66820.2025.11512536 |
| Palabra clave: | Data processing fault protection fault detection machine learning algorithms nearest neighbor methods phasor measurement units power system simulation support vector machine |