鲁棒预测迭代学习控制在间歇过程中的运用

Robust Predictive and Iterative Learning Control as Applied to Batch Process

考虑间歇反应中存在的非线性、实际情况中的输入输出约束要求和扰动的重复特性和非重复特性,将采用迭代学习和预测控制相结合的方法设计控制器,使得系统输出跟踪给定参考轨迹,最终使得间歇反应能够满足产品质量要求.由于迭代学习控制系统从本质上看汇聚了时间和批次两个变量,故可称为2维系统.针对2维系统,采用李亚普诺夫函数确保系统的稳定性并得到系统的控制序列,上述的控制序列可通过求解线性矩阵不等式求得.为了验证算法的有效性,将上述控制算法应用在对连续搅拌釜(CSTR)温度期望轨迹的跟踪控制中,仿真结果表明了控制算法的有效性.

Batch processes can be nonlinear,are prone to repetitive and non-repetitive disturbances,and often have input and output constraints.Iterative learning and predictive control are used to design a controller that makes system outputs track a given set-point profile.Now,despite non-linearities,disturbances,and con-straints,batch reactors can satisfy their production requirements.Iterative learning control systems contain two variables,time and batch,and so can be regarded,in essence,as two-dimensional systems.A Lyapunov function is used to stabilize these systems and LMIs （linear matrix inequalities）are used to optimize their control actions.A continuous stirred-tank reactor （CSTR）model,tracking a given reference temperature traj-ectory,isutilized to illustrate the effectiveness of the proposed control method.