Supervised Multi-Rejector of Periodic Disturbances for Nonlinear Systems Using Decoupled State Multimodel

What is it about?

In this paper, a multi-rejector of periodic disturbances is proposed for discrete-time nonlinear systems represented by a decoupled state multimodel. We report a decoupled state multimodel repetitive-predictive control based on a supervised algorithm to ensure reference trajectory tracking and periodic disturbances rejection. Partial predictors associated to the local controllers make the best choice of the most valid partial controller that meets the desired closed loop performances. The effectiveness of the supervised multi-rejector is shown via a simulation example. Obtained results are satisfactory and show a good rejection of periodic disturbances and reference trajectory tracking.

Featured Image

Why is it important?

Disturbances have adverse effects on the system behavior making the synthesis of the control system a difficult task. Indeed, several industrial systems are subject to external disturbances. Noting, for example, the power electronics converters and rotating machines which are the main sources of harmonic disturbances in electrical networks . In addition, the physiological movements caused by breathing and heartbeat are the sources of quasi-periodic disturbances during a robotic surgery. However, most physical systems are nonlinear. Sometimes, a satisfactory modeling solution can be found by a linearization approximate. But, the characterization of physical systems using a linear model based on simplifying assumptions may result degradation of the representative model performance. To overcome this problem, we propose a multimodel approach. This approach presents a powerful technique for nonlinear systems modeling and ensures a good compromise between accuracy and simplicity of representative models. In addition, it offers simple properties that can be exploited in the synthesis of the control law and which particularly promote the extension of some linear systems results without specific analysis on the non-linearity. This work deals with a supervised repetitive-predictive control for discrete-time nonlinear systems represented by decoupled state multimodel. This proposed control law is designed to track the reference trajectory and to reject periodic disturbances.