Developed tools for an optimization approach, application to optimize a flyback structure volume under EMC and loss constraints

Abstract

The paper deals with a methodology to size and optimize static converters using a gradient optimization algorithm. Generally, these static converters are described by semi-analytical models (both symbolic and numerical models). These models are built from stair functions (manufacturer data to evaluate the volume and the losses in the transformer) or from complex series to calculate the low frequency spectrum and the electromagnetic compatibility (EMC) spectrum, or from real series to calculate the semiconductor losses. That generates a strong discontinuity of the functions and their derivatives. To avoid these constraints, the authors have developed generic tools to carry out each kind of constraint with a gradient optimization algorithm. These tools are integrated in an optimization software and used to carry out an optimization process. The developed approach is applied to optimize the volume of a flyback structure in a power factor correction (PFC) mode, to respect EMC standards and to minimize the whole losses in this structure. The optimization results are validated thanks to a measurement workbench.