Polytopic black-box modeling of dc-dc converters

The objective of this work is to develop modular black-box converter models for system-level design and analysis that are valid in a wide variety of operating conditions. The approach used to address this problem is to divide the converter operating space into sub-spaces. For each sub-space, a linear local model is constructed from frequency response functions (FRFs) measured at the converter terminals. The FRFs are then post processed using system identification, the linear models obtained are valid as a local representation of the converter intrinsic dynamic at that operating condition. In the regions where two or more adjacent linear models overlap, an appropriate weighting is applied between them to produce an accurate approximation of the measured data. This new structure allows the characterization of converters with rich nonlinear dynamics, such as an unregulated dc-dc bus converter working near the boundary region between discontinuous and continuous conduction modes or a regulated flyback dc-dc converter. These two converters are effectively modeled with the proposed approach, and its models are presented in detail, providing an in-depth description of their implementation as well as their extensive experimental validation with laboratory prototypes.