Comparative Analysis of Random Forest and Gradient Boosting Algorithms in the Classification of Dissolved Gas Analysis Data

Power transformers are critical electrical machines in the process of transmitting and distributing electrical energy. It is crucial that electrical energy reaches consumers in a high-quality and sustainable manner. Within this production process, it is essential to perform complete maintenance on power transformers to reduce losses and prevent potential major failures, and to ensure that electricity transmission and distribution processes are efficiently delivered to consumers. In this context, dissolved gas analysis (DGA), one of the fault diagnosis methods for power transformers, emerges as the ideal solution. In the analysis of DGA data, many classic methods are used, and in recent years, machine learning algorithms have also played a significant role. In this study, the evaluation of transformer faults was conducted by analyzing real gas analysis data using Random Forest and Gradient Boosting algorithms, in addition to conventional methods. It has been observed that the evaluation results of machine learning algorithms are more positive compared to conventional methods.