A Novel Quadratic Programming Based Model-free Adaptive Control for I/O Constrained Nonlinear Systems

摘要

A novel quadratic programming (QP) based model-free adaptive control approach is proposed in this work to deal with constrained nonlinear discrete-time systems. Three types of constraints are considered together in this work, i.e., the constraints on the boundary of control input, the boundary of system output, as well as the change rate of control input between two time instants. All these constraints are transferred into a unified linear matrix inequality (LMI). Then, a control input index function is designed with respect to control errors and control input changes in a quadratic form. The control law is attained by minimizing the index function subjected to the LMI via quadratic programming (QP). The designed approach is data-based only and does not need an exactly linear model. Both theoretical and simulative results verify that the proposed approach is effective to applications. the effectiveness of the proposed approach.