Gray Box Modeling of Power Transformer Windings Based on Design Geometry and Particle Swarm Optimization Algorithm

Reliable models to predict the fast transient response of transformer windings are categorized into three groups: 1) white box; 2) black box; and 3) gray box models. Each group is applicable to solve a certain problem and includes its own advantages, hardly replacable by another group. The gray box methodology is a compromise between black and white box approaches which is investigated profoundly in this paper. A new method based on design geometry estimation is applied to accomplish transient simulation. The introduced procedure is probed on a noninterleaved double disk winding. Simulation and experimental results confirm that the gray box approach is applicable to both terminal and internal response estimation of a transformer when there is no knowledge of the transformer's internal specifications. The parameter identification of the model is performed by the particle swarm optimization algorithm.