Virtual prototyping and pre-sizing methodology for buck DC-DC converters using genetic algorithms

Abstract : In this article, the authors study the pre-sizing approach of DC-DC buck converters useful for electronics and system designers. The pre-sizing step in an industrial process is suitable to obtain a preliminary precise sizing of a sub-system (here a buck DC-DC converter): it permits the system designer to perform a feasibility analysis of an overall system (e.g. that contains a buck DC-DC converter). First, static modelling of power devices is proposed: MOSFET, diode, inductor and heatsink. According to an industrial context, the MOSFETs, diodes and heatsinks fitting curves are issued from the manufacturer-s datasheets. Second, the objective functions are explained in the case of mixed integer programming problems. Then, the optimisation variables and constraints are highlighted. Third, a section describes the choice of a multi-objective optimisation technique that leads to genetic algorithms (GAs). Fourth, the optimisation results are given. The choice of a final solution of the pre-sizing approach is discussed, considering additional constraints such as diode and MOSFET junctions temperatures, switching frequency etc. The authors focus on the general aspect of the optimisation method proposed here. It can also be used by power electronics designers with the help of additional constraints in accordance to their specific applications.
Document type :
Journal articles
Complete list of metadatas

https://hal-supelec.archives-ouvertes.fr/hal-00656896
Contributor : Stéphanie Douesnard <>
Submitted on : Thursday, January 5, 2012 - 1:53:50 PM
Last modification on : Monday, May 6, 2019 - 4:22:03 PM

Identifiers

Collections

Citation

Pierre Lefranc, Xavier Jannot, Philippe Dessante. Virtual prototyping and pre-sizing methodology for buck DC-DC converters using genetic algorithms. IET Power Electronics, The Institution of Engineering and Technology, 2012, 5 (1), pp. 41-52. ⟨10.1049/iet-pel.2010.0284⟩. ⟨hal-00656896⟩

Share

Metrics

Record views

171