Nonlinear model predictive control: specifying rates of exponential stability without terminal state constraints

摘要

A Model Predictive Control framework is proposed under which we can prescribe guaranteed rates of exponential stability for a class of nonlinear systems with input constraints. A feature of this framework is that the firxed-finite horizon open-loop optimal control problems (OCP) to be solved do not have any terminal state constraints imposed. This feature provides some important advantages: it significantly improves the efficiency of the optimization algorithm and more importantly the usual hypothesis on the existence of solution to the OCP (in general difficult to verify in the presence of terminal state constraints) is automatically satisfied. Another common hypothesis we are able to relax here is the continuity of the optimal controls. These facts combine to produce a framework in which the stability can be established under assumptions involving only the data of the nonlinear model and not involving the solutions to the optimal control problems.