A multi-parametric-programming-based feasible region-expansion algorithm for synthesis of explicit model predictive control system
基于多参数规划的显式模型预测系统设计的可行域扩张算法

Based on the multi-parametric quadratic programming(MPQP) for the linear constrained optimal control problem, a feasible and a feasible-region-expansion algorithm is proposed for synthesis of the explicit model predictive control(MPC) systems. Firstly, an iterative algorithm for computing the output invariant set of the optimal control problem is established; the computed invariant set is then used as the initial constraint for the state region in the MPQP problem. By iteratively solving the MPQP problem and replacing its constraint for the state region, the feasible region of the explicit MPC system is progressively expanded until its convergence. The state trajectory of the resulting explicit MPC systems starting from any state in its state partitions is always infinite time-feasible. Simulations also show that the proposed algorithm is effective.