Coordinate Descent Auto-Tuning Architecture for Multi Rail DC-DC Switch Mode Power Converters

Abstract

This paper proposes a new architecture for auto-tuning control of independent multi rail DC-DC power converters. The presented architecture is based on a low computational complexity parameter estimation and system identification technique using a Leading Coordinate Descent (CD) algorithm. A single system identification algorithm is used to estimate all the parameters of a multi rail DC-DC converter system and then determine optimal PID coefficients for each individual converter. An Equation Error (QE), Infinite Impulse Response (IIR) adaptive filter is used alongside the CD algorithm to model the Switch Mode Power Converter (SMPC). The presented results demonstrate the effectiveness and robustness of the proposed architecture, in terms of computation complexity, accurate estimation, and high performance dynamic response for all DC-DC converters present.