Distributionally Robust MPC for Nonlinear Systems

Abstract

Classical stochastic model predictive control (SMPC) methods assume that the true probability distribution of uncertainties in controlled systems is provided in advance. However, in real-world systems, only partial distribution information can be acquired for SMPC. The discrepancy between the true distribution and the distribution assumed can result in sub-optimality or even infeasibility of the system. To address this, we present a novel distributionally robust data-driven MPC scheme to control stochastic nonlinear systems. We use distributionally robust constraints to bound the violation of the expected state-constraints under process disturbance. Sequential linearization is performed at each sampling time to guarantee that the system's states comply with constraints with respect to the worst-case distribution within the Wasserstein ball centered at the discrete empirical probability distribution. Under this distributionally robust MPC scheme, control laws can be efficiently derived by solving a conic program. The competence of this scheme for disturbed nonlinear systems is demonstrated through two case studies.