Genetic algorithm optimization of peak current mode controlled buck conve

Abstract

This paper presents a genetic algorithm (GA), which optimizes the parameters of an analog controller for switched mode power converter (SMPC). A peak current mode controlled buck converter is used to test the optimization algorithm. The SMPC's response to line and load step changes is simulated with every combination of controller parameters emerging in GA's population. Each controller, i.e. each chromosome in the population, is assigned a cost depending on the simulated performance of the converter. The algorithm converges successfully. Although it relies on simulations, the measurements confirm that the controllers obtained by the GA result in a SMPC with stable and fast response with minimum over- and under-shoot. This method of controller optimization requires an accurate and reliable simulation model, but the transfer functions of the converter are not needed. Therefore, it can be most useful, if converter's continuous transfer function model is unknown, or if traditional controller design techniques do not yield satisfactory results.