Sparse Identification and SDRE based MRAC Scheme for a Van de Vusse Reactor. Issue 1 (2022)

• Record Type: Journal Article
• Authors: Xavier, Joanofarc; Patnaik, S.K.; Panda, R.C.
• Is Part Of: IFAC-PapersOnLine. Volume 55:Issue 1(2022)
• Page Start: 890
• Page End: 895
• Publication Date: 2022
• Subjects: SINDYc -- system identification -- nonlinear dynamics -- SDRE -- MRAC -- Van de Vusse reactor -- nonlinear; Automatic control -- Periodicals; 629.805
• DOI: 10.1016/j.ifacol.2022.04.146
• Languages: English
• ISSNs: 2405-8963

Abstract

Abstract: Recent powerful developments in the field of nonlinear system identification have a strong influence over nonparametric and machine learning based identification algorithms. In this paper, a sparse regression algorithm SINDYc (Sparse identification of nonlinear Dynamics with control) method is intended to investigate the dynamics of nonlinear physical systems. The SINDYc identification is applied to a nonlinear 'Van de Vusse reactor' and its performance is weighed against nonlinear least square identification method. Consequently, in this work, instead of using a linear reference model, SDRE (State dependant Ricatti equation) based nonlinear reference model is employed in the MRAC (Model reference adaptive control) for nonlinear physical systems. Furthermore, the closed loop-control performance of the reactor is investigated using SDRE based MRAC control. The Matlab simulations confirm that the proposed SDRE-MRAC control made a significant improvement (over other controllers) in the performance of closed loop control of Van de Vusse reactor in tracking the various set points and provided better control effort.