Multi-Objective Optimization of EE-core Transformers using Geometric Programming

This paper introduces a generalized optimization algorithm for the design of EE-core based transformers using geometric programming. The objective functions, loss and volume, and the constraints must be either posynomials (positive polynomials) or monomials (products of variables with exponents) to conform to the requirements of geometric programs. We can optimize transformers and generate a set of optimal solutions to construct a Pareto front based on inputs from a user. The algorithm, modelling and constraints are described. The output of the optimization has been compared to an experimental prototype to characterize the validity of the models.