A decision support system for online predictive maintenance of distribution transformers

Predictive maintenance of distribution transformers is generally considered a considerably challenging task, requiring the frequent inspection of specific transformer characteristics and sensor data. Nowadays, machine learning supervised methods constitute the state-of-the-art approach for leveraging these data for the development of intelligent computational detection systems. However, the lack of maintenance data exponentially increases the complexity of this problem. In this work, we propose a data-driven decision support system for online monitoring of the health state of a distribution transformer. An attractive advantage of the proposed approach is the fact that it does not require any prior knowledge about the transformer’s condition, addressing this problem of lack of labeled data. The reported analysis highlights the efficacy and the efficiency of the proposed approach, which reveals that can reasonably detect abnormal transformer behavior without the need of maintenance data.