A paper presented in ICECCME at Tenerefe
Dry power transformers are a critical component of microgrids, but their diagnostic can be challenging due to the various types of defects that can occur. This paper proposes several monitoring techniques to predict these defects and improve the diagnostic of dry power transformers in microgrids. One of the key features of this approach is the use of thermal image classification to detect the number of short circuits in the power transformer. The classification of thermal images is performed using the transfer learning method, which allows for the utilization of pre-trained models and the adaptation of them to the specific task at hand. This feature is integrated into a larger smart energy management system for microgrids, which aims to optimize the operation and maintenance of these systems. The proposed techniques have been tested and validated through experiments, and the results demonstrate their effectiveness in accurately predicting defects and improving the diagnostic of dry power transformers in microgrids.