A self tuning regulator design for nonlinear time varying systems based on evolving linear models

Abstract

Fostered by the current notion in the field of intelligent control systems which demands majority of controllers to be self-tuning, adaptive to parameters and structure changes, and above all, intelligent in the face of new circumstances, in this paper, for a general class of Single-Input Single-Output (SISO) nonlinear time-varying systems, a novel Self Tuning Regulator (STR) design based on an introduced Evolving Linear Model (ELM) is proposed. Under definite assumptions and specific constraints, even if there exists no priori knowledge about the system dynamics except its order and relative degree, a suggested online linearization technique based on Recursive Least Squares (RLS) method is applied in order to identify the plant and to derive an Adaptive Linear Regression (ALR) model for the system. An ALR could be treated as ELM when the number of independent regressors which construct the model varies over time. It is demonstrated that under certain constraints, SISO nonlinear systems could be represented by ELMs. Afterwards, an indirect STR strategy is explained and applied on the online identified ELM of the nonlinear system. Multifarious simulations were performed and results clearly demonstrated the privilege and effectiveness of the proposed approaches.

from #MedicinebyAlexandrosSfakianakis via xlomafota13 on Inoreader http://ift.tt/2b3uyeG
via IFTTT