Abstract:
Due to their several advantages, tape-wound core transformers are used in various power electronic systems. To obtain the optimal designs of the transformer using multi-objective optimization techniques, the effect of its performance on the whole system should be accounted for without computational liability. This may be achieved by deriving a meta-model based on scaling laws which relate the transformer performance equations to general quantities such as rated power, frequency and current density. A perunit T-equivalent circuit and magnetic equivalent circuit model are used to obtain the transformer scaled model. By using geneticalgorithms based multi-objective optimization and curve fitting techniques, the transformer meta-model is derived. The derived meta-model is used to study the effect of varying frequency and rated power on the transformer performance.