ARNISMC for MLS with global positioning tracking control

• Published in: IET electric power applications 2018-04, Vol.12 (4), p.518-526
• Main Authors: Chen, Seng-Chi, Kuo, Chun-Yi
• Format: Article
• Language: English
• Subjects: adaptive recurrent neural network; AR radial basis function NN estimator; ARNISMC; control system synthesis; global positioning tracking control; magnetic levitation; magnetic levitation system; MLS; MLS controller design; neurocontrollers; NN intelligent sliding‐mode controller; nonlinear dynamic model; nonlinear dynamical systems; recurrent neural nets; Research Article; SM control; variable structure systems
• ISSN: 1751-8660; 1751-8679; 1751-8679
• DOI: 10.1049/iet-epa.2017.0690

This study develops an adaptive recurrent neural network (NN) intelligent sliding-mode controller (ARNISMC) for magnetic levitation system (MLS). First, a non-linear dynamic model of the MLS is derived. Thereafter, a SM controller (SMC) method is presented to compensate for the uncertainties in the MLS. In addition, to enhance the control effort of a conventional SMC and further increase the tracking performance of the MLS, the uncertainty terms of the system dynamics can be estimated online by using an AR radial basis function NN estimator. Accordingly, the proposed controller not only offers the accurate positioning tracking control, minimises steady-state error, and improves conventional controller performance but also provides global positioning tracking control, which has not been previously reported in the literatures. It can effectively solve problems associated with typical MLS controller design. Moreover, the satisfactory tracking performance can be observed from the experimental results by adopting the proposed ARNISMC scheme.