A Supervised Classification Method Based on Logistic Regression with Elastic-net Penalization for Heat Waves Identification to Enhance Resilience Planning in Electrical Power Distribution Grids
Résumé
In the last years, extreme weather events, including heat waves, have caused extensive and long-lasting interruptions in the electric power distribution systems of large urban areas. According to the current Performance-Based Regulation (PBR) approaches for incentives aimed at strengthening utility performance, these interruptions have to be quantitatively measured in network resilience metrics, rather than reliability metrics. In this distinction between resilience and reliability, a clear definition of the extreme events is required. This, however, is still lacking. To address the problem, in this work we propose a method to define heat waves, which relies on logistic regression with elastic-net penalization to quantitatively associate environmental and operating conditions of the network to significant increments of its failures. The methodology is validated by application to the medium voltage distribution network of the city of Milano, Italy.