A new approach to fast simulation of periodically switching power converters

Abstract

Periodically switching power converters can be described by an averaged equation and ripple correction functions. An algorithm for the analytical computation of the averaged system and the ripple correction functions, the Krylov-Bogoliubov-Miltropolsky algorithm is implemented using a symbolic computer program. A ripple correction series, which estimates the actual waveforms given the averaged waveforms, is derived. Correction terms for the averaged model are also derived. The approximations require numerical calculations of averaged models which are normally not stiff. The correction terms are continuous functions which can be computed symbolically. The approximate system can be integrated on time scales slower than those used for conventional time-domain techniques. A resonant inverter and a boost converter are presented as examples.<>