Performance evaluation of an autoencoder state estimator with realistic low voltage grids reconstructed from open data
Résumé
Distribution systems operators' (DSO) upcoming challenge is to monitor and control low voltage (LV) grids. Real time measurements' acquisition from LV grids has become possible thanks to the deployment of smart meters (SM) but it is still limited by technical constraints. Thus, a real time state estimator (SE) is needed to ensure the LV grid's observability. A machine learning model can be trained using the SM historical data and then run in real time using few available measurements to estimate the system's state. For the purpose of evaluating the performance of a machine learning based SE on a realistic LV grid, an auto-encoder based SE is presented in this paper. An accuracy enhancement using pseudomeasurements is proposed and a method for reconstructing realistic LV grid models is also described.
Fichier principal
PERFORMANCE EVALUATION OF AN AUTOENCODER STATE ESTIMATOR WITH REALISTIC LOW VOLTAGE GRIDS RECONSTRUCTED FROM OPEN DATA.pdf (710.83 Ko)
Télécharger le fichier
| Origine | Fichiers produits par l'(les) auteur(s) |
|---|