工业过程控制系统迭代学习算法的增益确定

应用频域和时域（非状态空间法）相结合的方法对简单工业过程控制系统迭代学习算法进行了收敛性分析，在频域得出了用系统参数显示表示的收敛性条件，避免了收敛条件的验证对系统时域模型参数的依赖性，使验证更简洁，用平方积分鉴定法确定了首次学习时误差平方积分最小意义下学习增益的最优值，明确了学习增益选取的目标，数字仿真表明：所确定的学习增益不仅是最优的，而且相应的迭代学习控制能显著改善控制系统的动态品质。

Determination of the Iterative Learning Gains for Industrial Process Control Systems

The convergence of the iterative learning control algorithms for industrial process control systems is analyzed by combining frequency domain with time domain methods, and the convergence condition is explicitly expressed by the model based parameters of the system in frequency domain. Hence the verification of convergent condition is more simple since its dependence on the parameters in time domain is avoided. The optimum of the learning gains is determined in the sense that the integral of square error of the first learning is minimal, thus the aim of choosing learning gains is clear. The digital simulations indicate that not only the determined learning gains are optimal but also the corresponding iterative learning control can remarkably improve the dynamic performance of the control system.