An Offline Optimization and Online Table Lookup Strategy of Two-layer Model Predictive Control

Two-layer model predictive control is restricted in the field of high real-time control and poor computational controller, due to the high computational complexity. In this paper, we propose an offline optimization and online table look-up strategy of two-layer model predictive control to address the problem. In the upper layer of steady-state target calculation, offline optimization and online table look-up is developed to calculate steady-state targets. In the lower layer of dynamic optimization, an unconstrained model predictive control is adopted to track the steady-state targets from the upper layer. For the case that online table look-up method is infeasible, multi-parametric LP and linear weighted sum methods are presented. An exhaustive simulation of a fat system is made to explore the performance of the strategy of two-layer model predictive control. Furthermore, the properties of computation complexity, steady-state and robustness are examined from the point of implementation. Simulated studies demonstrate the effectiveness of the proposed strategy.