CLASS-T: Automatic classification of Transient in Electrical network

I am Principal Investigator of this project. Its purpose is to define and to test in real time condition an artificial intelligence (AI) system for the classification of different sources of transients in power networks (partial discharges, corona effect, loads…). The approach is defined as two stages AI: one distributed in the network and another one – central – merging the information classified locally. Our current results show an interest of a hybrid AI system, machine learning-based at distributed level and deep learning at the central level. Tests in real configurations are currently conducted thanks to the collaboration with ENEDIS, Lyon.

The project is founded by Auvergne-Rhone-Alpes Region (Pack Ambition Recherche program) - 2019-2023
Thanks to Auvergne Rhone Alpes region support, many publications have been done:
  • A Novel Approach for Characterization of Transient Signals Using the Phase Diagram Features - Download .PDF
  • Entropy-based characterization of the transient phenomena – systemic approach - Download .PDF
  • On the existing and new potential methods for Partial Discharge source monitoring in electrical power grids - Download .PDF