Robust observer-based model predictive control of non-uniformly sampled systems

This paper presents a robust model predictive control design procedure for constrained non-uniformly sampled systems subject to bounded unknown disturbances. Unlike the existing results in this area, the proposed design procedure considers arbitrary variations of the sampling period between a lower and an upper bound, and ensures asymptotic stability and performance of the closed-loop control system for all variations of the sampling period, which is a theoretically challenging task. The uncertain optimization problem associated with the constrained model predictive control turns out to be intractable due to infinity many variations of the sampling period. To overcome this challenge, the non-uniformly sampled system is modeled as a linear polytopic uncertain system based on variations in the sampling period. A quasi min-max robust model predictive control technique based on an offline state observer is used for designing an output feedback controller. A case study is given to illustrate the effectiveness of the proposed approach.