Detection and Analysis of Faults in Transformer using Machine Learning

Transformers perform a very important role in the power grid. The distribution transformer is the most pivotal part and its failure may result in great economic loss and interruption in the power supply. However, prediction of fault in distribution transformers has its own limitations such as binary classification effect, low data quality, and small sample learning. Magnetic oil gauge fault may cause major problem in the distribution transformer, thus prediction of Magnetic Oil Gauge fault with the help of other parameters that may cause the fault will save a lot of resources and money. This paper proposes a method where this can be achieved. Transformer parameter like Oil Temperature Indicator, Winding Temperature Indicator, Magnetic Oil Indicator, is visualized using python libraries. Models like logistic regression, Support Vector Machines, K-Neighbors Classifier, Decision Tree Classifier, Random Forest Classifier, Ada Boost Classifier, XGB Classifier are implemented. Current values from the data set is used to implement an overcurrent relay. The occurrence of the fault is predicted using the data of the Magnetic Oil Gauge. The discussion of the results of experiments is provided with the comparative study of the different models.