基于线性集结的预测控制器

Predictive controller based on linear aggregation

针对一类受约束的线性离散系统,研究了基于集结预测控制器的可行性问题．通过一个列满秩的集结矩阵将维数较小的控制变量序列映射成在线优化变量序列,在不缩短控制时域的情况下,降低了在线优化变量的个数．给出了优化变量可集结的充分条件,当集结矩阵满足这个条件时,在线求解优化问题总会得到可行解．最后,给出了一个仿真实例,仿真结果很好地验证了本文的结论．

For a class of constrained linear discrete-time systems, the feasibility problem of aggregation-based predictive controllers is investigated. Using an aggregation matrix with full column rank to map the on-line optimization variable sequence into a new one, the number of on-line optimization variables is cut down without reducing the control horizon. A sufficient condition is presented to check whether the control variable sequence can be aggregated. When the selected aggregation matrix satisfies this condition, a feasible solution can always be obtained by solving the on-line optimization problem, Finally, a simulation example is given to illustrate the proposed result.