Optimal Load Transient Response of the Boost DC-DC Converter Based on Stochastic Duty-cycle Sequence Generator

This paper presents simplified method allowing to obtain time optimal sequence of the duty-cycles, corresponding to the best possible transient response of the boost DC-DC converter to a general load transient pattern. It aims to evaluate (benchmark) practical limits of the regulation loop, and margins of the feedback-loop improvement. Optimal transient response allows also to generate training sequences for machine learning ARX-based controllers. Generation of the optimal sequence is based on randomized generation, evaluated by using integral quality criterion as penalty function. This simplified method present advantage against analytical methods (e.g. based on Pontryagin's minimum principle) allowing easy customizing. Results are analyzed on example of ideal load current step. Optimal duty-cycle sequence is described as a mathematical function of a load step current amplitude. Demonstration showing comparison of a traditional analog peak current mode boost converter with optimal duty-cycle sequence is presented, providing more than 50% reduction of recovery time. Target of this comparison is optimization of the feedback controller for general-purpose and automotive boost dc-dc converters.