Machine Learning Applied to Shutdown Risk Estimation in Transmission System Maintenance Interventions

The implementation of machine learning algorithms has become present in recent years in the electric sector. The increasing size of the system and its complexity, coupled with obtaining the large amounts of collected data enabled machine learning to become an important and effective tool for power systems operation. This paper presents a machine learning based algorithm for the estimation of shutdown risk during the execution of services in transmission system equipment. The algorithm extracts knowledge about shutdown risk from the Brazilian Independent System Operator (ONS - Portuguese acronym) maintenance database, supporting the analysis of the intervention requests from the transmission agents. A simple text mining process is conducted aiming to obtain relevant features to the shutdown risk estimation model. The results show that a simple machine learning algorithm can extract useful information from an unstructured database, helping the operational teams in the transmission intervention requests analysis process.