Power Transformer Characterization and Design Optimization Environment

Power Transformers are essential part of all AC power grids. This work proposes a characterization and design optimization environment for power transformers to increase efficiency while decreasing core volume. It utilizes finite-element state-space (FE-SS) models in conjunction with artificial neural networks (ANNs) and particle swarm optimization (PSO). The results of a case study involving a 42 MVA, 118/13.8 kV distribution step down power transformer are presented, where the transformer performance characteristics are computed and compared to test data for verification. Next, the optimization environment is used to decrease the transformer volume and improve its efficiency. In addition, the Taguchi orthogonal arrays (OA) method is used to reduce computational costs.