Model Predictive Control for a Multicompartment Respiratory System

In this brief, we develop a model predictive controller for a multicompartment respiratory system. Specifically, for a given periodic reference volume pattern, we design a tracking controller using a framework that merges repetitive control and model predictive control. In particular, the periodic multicompartment model is transformed into a lifted run-to-run invariant model, and a model predictive controller with nonnegative control input constraints is designed. The proposed tracking control framework is applied to a two-compartment lung mechanics model to demonstrate the efficacy of the proposed approach.