CLASSIFICATION OF LEAKAGE REACTANCE IN TWO-LAYER CYLINDRICAL-WOUND TRANSFORMERS USED IN ZIGZAG CONNECTING USING ENSEMBLE BAGGED TREES (EBT)

Leakage reactance plays a crucial role in the performance and efficiency of transformers.In particular, two-layer cylindrical-wound transformers with zigzag connections exhibit complex leakage reactance behaviors due to their interlaced winding structure.Traditional methods for determining leakage reactance rely on analytical modeling and finite element simulations, which are often computationally expensive.This study proposes a machine learning-based approach utilizing the Ensemble Bagged Trees (EBT) model to classify leakage reactance levels based on key transformer design parameters.A dataset comprising primary winding turns, coil radius, insulation thickness, and Rogowski coefficient was used to train and validate the model.The experimental results demonstrate that the EBT model achieves 86.25% classification accuracy, proving to be a reliable and computationally efficient alternative for predicting leakage reactance.The study also highlights the most influential design parameters affecting leakage reactance, providing valuable insights for transformer optimization.Future work will focus on incorporating deep learning techniques to further enhance classification performance and exploring real-world validation scenarios.