A Novel Adaptive LSSVR-Based Inverse Optimal Controller With Integrator for Nonlinear Non-Affine Systems


Creative Commons License

Sancı M. E., Ucak K., Gunel G. O.

IEEE Access, vol.11, pp.137548-137576, 2023 (SCI-Expanded) identifier

  • Publication Type: Article / Article
  • Volume: 11
  • Publication Date: 2023
  • Doi Number: 10.1109/access.2023.3339571
  • Journal Name: IEEE Access
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus, Compendex, INSPEC, Directory of Open Access Journals
  • Page Numbers: pp.137548-137576
  • Keywords: Adaptive control, inverse optimal control, NARMA-L2 model, online LSSVR
  • Istanbul Technical University Affiliated: Yes

Abstract

In this study, a novel inverse optimal controller based on NARMA-L2 modelling technique and online least squares support vector regression (LSSVR) method has been proposed for nonlinear non-affine systems. Firstly, the nonlinear autoregressive with exogenous inputs (NARX) model of the system is obtained using online LSSVR method, then this model is decomposed into NARMA-L2 submodels. Hence, the non-affine system model is converted to a nonlinear affine system model. The obtained NARMA-L2 submodels are used in computing the inverse optimal control law. Furthermore, the parameters of the inverse optimal controller have also been optimized online using the Levenberg-Marquadt algorithm. The performance of the proposed LSSVR based inverse optimal controller using NARMA-L2 model has been evaluated by simulations carried out on two benchmark systems, and the results show that the LSSVR based NARMA-L2 model and inverse optimal controller attain good modelling and control performances.