Power transformer data for fault diagnosis

Power transformers are essential machines in every electricity generation and distribution grid. Its reliability must be assured as an unexpected failure might represent a catastrophic event leading to a lack of the electricity ser- vice in a whole community or region. Given the importance of such trans-formers, predictive maintenance techniques have been increasingly used to detect incipient faults and trigger maintenance actions accurately when are needed. One of the traditional analyses on power transformers is on oil gas analysis. Multiple methods have been proposed to detect and identify faults on different variables extracted from the oil gas analysis. However, these traditional techniques not always offer a conclusive diagnostic. Given the in- creasing use of data-driven models in fault diagnosis, this study was focused to collect and organize a data from oil gas analysis preformed on real transformers over multiple years of service. The objective is to provide a dataset that researchers can use to train innovative data-driven models that might overcome the current gaps in power transformers fault diagnosis using oil gas data.