Predicting the Uncertainty of Data Generated by Power Transformer using Gradient Boosting Decision Tree

Power distribution and transmission depend heavily on transformers. Transformers encounter a variety of problems, including low voltage power, power loss, power wastage, power theft, phase-to-phase faults, overloading, and others. The HI (also known as health index) tool is used to address power transformer issues. The method of calculating the power transformer’s health index is challenging and time-consuming, and there is a greater probability of data uncertainty and missing data. With the help of the Gradient boost decision tree(GBDT) method, this research intends to increase the accuracy of determining the health status of the power transformer. GBDT is a machine learning method for improving the prediction power of the model across successive learning stages. The GBDT is compared with Random forest (RF) and the accuracy provided by GBDT is 97.5%.