This paper develops a data-driven learning framework for approximating the feasible region and invariant set of a nonlinear system under the nonlinear Model Predictive Control (MPC) scheme. The developed approach is based on the feasibility information of a point-wise data set using low-discrepancy sequence. Using kernel-based Support Vector Machine (SVM) learning, we construct outer and inner approximations of the boundary of the feasible region and then, obtain the feasible region of MPC for the system. Furthermore, we extend our approach to the perturbed nonlinear systems using set-theoretic method. Finally, an illustrative numerical example is provided to show the effectiveness of the proposed approach.
@article{arxiv.2012.03428,
title = {Data-driven approximation for feasible regions in nonlinear model predictive control},
author = {Yuanqiang Zhou and Dewei Li and Yugeng Xi and Yunwen Xu},
journal= {arXiv preprint arXiv:2012.03428},
year = {2020}
}