基于输入输出线性化的连续系统非线性模型预测控制

非线性约束预测控制关键是求得可行性优化解. 输入输出反馈线性化是非线性控制一种常用的方法, 其系统的初始线性输入约束转化成非线性基于状态的约束, 因而无法采用常规的二次规划(QP)求解优化问题. 针对连续状态空间模型系统, 本文提出迭代二次规划方法来寻求非线性优化解. 为了保证算法的收敛性, 系统加入另外一种迭代算法来保证其在整个预测时域上能得到可行解. 仿真控制结果表明了该方法的有效性.

Continuous-time nonlinear model predictive control with input/output linearization

Obtaining a feasible optimal solution is the key issue in designing the constrained nonlinear predictive control. The input/output feedback linearization is a conventional method for this purpose. Because the original linear input constraints are converted to nonlinear and state-dependent constraints in this method, it makes the quadratic programming (QP) method inapplicable to obtain an optimal solution. To find the optimal solution for a state-space continuous-time system, we present an iterative quadratic program (QP) method. To guarantee its convergence, another iterative approach that can guarantee a feasible solution over the entire prediction horizon is incorporated. Simulation results of application to a continuous stirred tank reactor (CSTR) demonstrate the effectiveness of the proposed method.