Proactive power outage impact adjustments via machine learning

Abstract

A system and method for adjusting estimated number of customers affected by an electrical power distribution network incident. Incident reports indicating a number of affected customers and an incident definition for incidents occurring over an accumulation time are accumulated, where each incident report. For each incident report, a signature is determined based on characterizing respective incidents and includes respective characteristics of the incidents associated with each incident report. A subject incident report is received subsequent to the accumulation time. A signature is determined for the subject incident report. Similar incident reports that have signatures similar to the subject incident are identified. A composite estimated number of affected customers for the subject incident is determined based on the associated numbers of affected customers for each similar incident report. An initial estimate of customers affected by the subject incident is adjusted based on the composite estimated number of affected customers.