Iterative-learning control of linear continuous-time systems with disturbances (Q1596533)

An algorithm for iterative-learning control is presented for linear continuous-time MIMO systems \(\dot x= Ax+ Bu+\eta\), \(y= Cx+\sigma\), \(x\in\mathbb{R}^n\), \(u\in \mathbb{R}^m\), \(y\in \mathbb{R}^p\), where signals \(\eta\in \mathbb{R}^n\), \(\sigma\in\mathbb{R}^p\) can model disturbances or system uncertainties, and \(\sigma(t)\) is a \(C^1\) function. The proposed control law has the form \(u(t,k)= u(t, k-1)+ K\dot e(t, k-1)\), where \(e(t, k)= y_r(t)- y(t,k)\) defines the tracking error with respect to the reference trajectory \(y_r(t)\in C^1\). It is shown that such a control law exists if and only if \(\text{rank }CB= p\) and it is convergent if and only if the matrix \(\Phi= I-CBK\) has all eigenvalues inside the unit circle. For the particular case of controllable and observable SISO systems, an iterative-learning control law can be developed if \(y_r(t)\), \(\eta(t)\), \(\sigma(t)\) are \(C^n\) functions.