论文信息 - State dependent parametrizations for nonlinear MPC

State dependent parametrizations for nonlinear MPC

Abstract This paper aims at reducing the computational load caused by the online optimization in nonlinear model predictive control (MPC) by introducing a new type of parametrizations for the predicted input trajectory. In order to determine offline the state dependent parametrizations, a tailored data mining algorithm is introduced. Refinements to achieve feasibility of the parametrized constrained optimization problem are presented. Theoretical guarantees on constraint satisfaction and feasibility employing the parametrizations are provided for Lipschitz continuous systems. In a numerical example the benefits of the new method are illustrated.

Frank Allgöwer | Gregor Goebel | F. Allgöwer | G. Goebel

[1] S. Joe Qin,et al. A survey of industrial model predictive control technology , 2003 .

[2] K. Gabriel,et al. Least Squares Approximation of Matrices by Additive and Multiplicative Models , 1978 .

[3] Martin Mönnigmann,et al. Low complexity suboptimal explicit NMPC , 2012 .

[4] Chih-Jen Lin,et al. A comparison of methods for multiclass support vector machines , 2002, IEEE Trans. Neural Networks.

[5] Frank Allgöwer,et al. Improved state dependent parametrizations including a piecewise linear feedback for constrained linear MPC , 2014, 2014 American Control Conference.

[6] Frank Allgöwer,et al. Obtaining and employing state dependent parametrizations of prespecified complexity in constrained MPC , 2013, 52nd IEEE Conference on Decision and Control.

[7] F. Allgöwer,et al. A quasi-infinite horizon nonlinear model predictive control scheme with guaranteed stability , 1997 .

[8] David Q. Mayne,et al. Constrained model predictive control: Stability and optimality , 2000, Autom..