Predictive maintenance of power transformers based on a data-driven application

This paper presents an innovative digital tool called "Asset Management Optimization for Power Grids Application" based on Artificial Intelligence techniques for monitoring critical assets in power systems, such as power transformers. This application implements different features that target the operators of transmission and distribution systems, aiming at providing analytics and forecast models that optimize the management of maintenance activities. Unlike the traditional approaches estimating the conditions of critical transformers mainly through Dissolved Gas Analysis, the proposed application is based on low-cost and easily accessible data, typically from SCADA, to estimate the probability of failure. The article concludes by focusing on one pilot installation in a distribution network for preliminary testing purposes in a real-life environment.