On Min-Max Robust Data-Driven Predictive Control Considering Non-Unique Solutions to Behavioral Representation

Abstract: Direct data-driven control methods are known to be vulnerable to uncertainty in stochastic systems. In this paper, we propose a new robust data-driven predictive control (DDPC) framework to tackle the uncertainty in dynamic systems. By analyzing non-unique solutions to behavioral representation, we first shed light on the lack of robustness in subspace predictive control (SPC) as well as the projection-based regularized DDPC. This inspires us to construct an uncertainty set that includes admissible output trajectories deviating to some extent from nominal predictions from the subspace predictor and develop a min-max robust formulation of DDPC that endows control sequences with robustness against such unknown deviations. We establish its performance guarantees under bounded additive noise and develop convex reformulations of the min-max problem with moderate complexity. To mitigate the conservatism of robust design, an adaptive robust DDPC scheme is further proposed by incorporating an affine feedback policy, with performance guarantees and tractable reformulations derived. Simulation studies show that the proposed method effectively robustifies SPC and outperforms projection-based regularizer.