Generation and Derivation of Practical Optimization-Oriented Models of Inductors

Magnetic components contribute significantly to the volume and losses of power electronic systems. Their optimal design is therefore crucial for the overall optimization of power density and efficiency of power converters. Recently it has been shown that converters can accurately be modeled using posynomial functions, thus allowing for the use of Geometric Programming, a type of convex optimization problem, to be used to quickly produce globally optimum designs of entire converter families. Existing studies have however treated magnetic components in a case-specific way, deriving posynomial models suited for the given converter or range of operating points. This paper demonstrates and validates methods for the derivation and generation of posynomial models of inductors for an entire catalogue of standard components, allowing a set of models to be generated once and then re-used subsequently, or "plugged into" an overall converter optimization, repeatedly, without the need for re-derivation.