Disturbance-adaptive Model Predictive Control for Bounded Average Constraint Violations

This paper considers stochastic linear time-invariant systems subject to constraints on the average number of state-constraint violations over time without knowing the disturbance distribution. We present a novel disturbance-adaptive model predictive control (DAD-MPC) framework, which adjusts the disturbance model based on measured constraint violations. Using a robust invariance method, DAD-MPC ensures recursive feasibility and guarantees asymptotic or robust bounds on average constraint violations. Additionally, the bounds hold even with an inaccurate disturbance model, which allows for data-driven disturbance quantification methods to be used, such as conformal prediction. Simulation results demonstrate that the proposed approach reduces closed-loop cumulative cost compared to state-of-the-art methods across different target violation rates, while satisfying average violation bounds.