二次型鲁棒迭代学习控制在间歇过程中的应用

Application of robust iterative learning control with quadratic performance for batch process

主要研究二次型迭代学习控制算法在间隙反应器中的应用,旨在解决系统输出对系统期望轨迹的跟踪问题。针对非线性间歇反应器,文中采用系统输出在标称轨迹输入下线性化的时变偏扰模型进行描述,时变偏扰模型通过最小二乘法辨识得到。同时考虑系统正常运行时往往会出现不确定性,故将迭代学习算法和S-procedure鲁棒方法相结合,对时变偏扰模型设计鲁棒迭代控制算法,并将问题转化为线性矩阵不等式可行性求解问题。仿真结果表明在参数发生变化时系统仍有较强的鲁棒性,能有效跟踪期望轨迹。

This paper presents an iterative learning control with quadratic performance （Q-ILC） applied to batch process and aims at solving the problem of tracking the output desired trajectory in batch process. For nonlinear batch process, time-varying perturbation models, linearized around the nominal trajectories, are identified from the operating data using the least squares methods. Model uncertainties always exist during the process, so it can combine the Q-ILC and the analysis of S-procedure to design the robust ILC for time-varying perturbation models. Then, the problem can be reduced to a feasibility problem of linear matrix inequalities （LMIs）. Simulation results demonstrate the robustness of uncertain system and the system output variable can track the expectation trajectory effectively.