System identification of power converters based on a black-box approach

Abstract

This paper addresses the topic of a model-free system identification of highly nonlinear power electronics systems from the data either through a time-domain simulation or a hardware measurement. This system identification method based on the black-box approach using the data generated from well-known design tools such as PSpice and MATLAB is generally simpler and independent of type of converter. As an application of identifying an unknown plant in power electronics systems, a constructive black-box approach is presented which aims at generating discrete-time small-signal linear equivalent models for a general class of converters, which includes resonant converters, pulsewidth modulation (PWM) converters and zero-voltage-switched (ZVS) PWM converters. The resulting small-signal model describes the converter as a linear time-invariant system, and the knowledge of the identified linear system can be applied to the switching converters for constructing feedback controllers. The identification results are compared with the analytical model and experimental data.