A Kind of Model Predictive Control under Model Mismatch

On the basis of the equivalent dynamic linearization model (EDLM), we propose a kind of model predictive control (MPC) for single input and single output (SISO) linear systems. Practically, when MPC is designed tier the situation of model mismatch, the actual system performance may beyond our expectation. This paper is concerned with the system performance analysis of the proposed method and explains how it works. Most importantly, fewer works are able to analyze the system transient characteristics and steady-state characteristics more clearly than this work, when the system model is offline built inaccurately or the model parameters are online estimated imprecisely. In this paper, we formulate the system characteristics through analyzing the closed-loop function of the system when the system model is inconsistent with the real system, and then we deduce several useful conclusions about the system performance. By this manner we can understand the principle of the controller and know the relationship among the controller, system model and the system characters. This may help us to know how to design the controller to achieve our desired system characteristics. At last, simulations are carried out to verify the conclusions.