Nonlinear predictive control

Abstract

The dissertation deals with the design of a discrete predictive controller on a finite horizon for general nonlinear systems described by state equations, which are based on the knowledge of a continuous nonlinear control system model. The proposed nonlinear predictive controller is a modification of the design of a predictive controller for a discrete linear MIMO state control system model and a matrix criterion in full deviation form. The implementation of a general controller design for two different nonlinear systems is shown – the „Lotka-Volterra“ system taken from the literature, supplemented with control inputs and a real laboratory device „magnetic levitation“. The behavior of the proposed controllers is shown in the simulations of control processes in the MATLAB environment. In the case of magnetic levitation, the work also deals with the issue of obtaining a mathematical model of the laboratory device and estimating the unmeasured state of the nonlinear system. Real-time control is also demonstrated for this system.