Plant-model mismatch evaluation for unconstrained MPC with state estimation

摘要

In this paper, we develop an autocovariance-based method for estimating plant-model mismatch in unconstrained linear model predictive control systems that use a Kalman filter state estimator. Assuming knowledge of the noise structure, we derive an expression for the autocovariance of the process outputs as a function of (an additive) plant-model mismatch. We then formulate the mismatch estimation problem as an optimization problem aiming to minimize the discrepancy between the theoretical autocovariance and the sample estimator of the autocovariance, obtained from closed-loop operating data collected during steady-state operation. Case studies are provided to demonstrate the performance of the approach.