Use of evolutionary algorithms for design optimization of power converters
Authors:
- 1 February 2012
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 268-272
Abstract
This paper presents an optimal design approach of power converters by using a multi-objective formulation with discrete variables in which three conflicting objective functions are defined to perform an optimum selection of power components: minimize area, minimize cost, and maximize efficiency. NSGA-II algorithm is selected among several multi-objective combinatorial optimization methods suitable to handle discrete variables (i.e. MOSFETs, inductors, capacitors). Proposed methodology is applied to a synchronous buck converter taking into account the current load profile analyzed with the Probability Distribution Function to identify the different load power states. Power losses are evaluated for each of the power states resulting in a weighted function that accounts for the different current values under different usage cases. NSGA-II algorithm is used to find the Pareto front of a very large solution space, with hundreds of thousands feasible combinations, that allows the designers to make the right choice based on a trade-off analysis when selecting the power components.Keywords
This publication has 5 references indexed in Scilit:
- Design of an 99%-efficient, 5kW, phase-shift PWM DC-DC converter for telecom applications2010
- Power delivery for the next generation mobile platform2008
- A study on Load Power Probability-Distribution-Function effect on power conversion efficiency2008
- A fast and elitist multiobjective genetic algorithm: NSGA-IIIEEE Transactions on Evolutionary Computation, 2002
- Algorithms for Power Converter Design OptimizationIEEE Transactions on Aerospace and Electronic Systems, 1981