Offline Robust Model Predictive Control for Nonlinear Uncertain Systems

Shizhong Yang,Lijuan Xing
DOI: https://doi.org/10.1109/cac.2017.8242778
2017-01-01
Abstract:In order to reduce the online calculation amount and improve the response speed, this paper proposes a robust model predictive control algorithm for nonlinear uncertain systems based on linear matrix inequality (LMI). The algorithm is divided into two parts: offline algorithm and online algorithm. In the offline algorithm, a sequence of asymptotically stable invariant ellipsoids and the corresponding control laws are got by one dimensional state variables optimization. The paper gives the calculating formula for one dimensional state variables parameter values, and explains the control laws got by the offline algorithm can guarantee the rapidity and stability of the system control. Due to the offline algorithm has completed most of the computation, when online, It only need to find suitable control laws in the parameter table according to the position of the system state variables in the ellipsoids, so it reduces the online algorithm complexity and the calculation cost. Through the simulation control effect comparing between the new algorithm and the traditional online control algorithm, the simulated results demonstrate the effectiveness of the new algorithm.
What problem does this paper attempt to address?