Sampling-based Stochastic Data-driven Predictive Control under Data Uncertainty
We present a stochastic output-feedback data-driven predictive control scheme for linear time-invariant systems subject to bounded additive disturbances and probabilistic chance constraints. The approach uses data-driven predictors based on an extension of Willems' fundamental lemma from behavioral systems theory and a single persistently exciting input-output data trajectory. Compared to current state-of-the-art approaches that rely on availability of exact disturbance data, we deterministically approximate the chance constraints in a sampling-based fashion by leveraging a novel parameterization of the unknown disturbance data trajectory, considering consistency with the measured data and the system class. A robust constraint on the first predicted step guarantees recursive feasibility of the proposed controller as well as constraint satisfaction in closed-loop. We show robust asymptotic stability in expectation under further standard assumptions. A numerical example demonstrates the efficiency of the proposed control scheme.
PDF Abstract
