LMI based stability analysis and robust controller design for discrete linear repetitive processes (Q4452311)

Discrete linear repetitive processes are considered, having the state-space representation \[ x_{k+1}(p+1)=Ax_{k+1}(p)+Bu_{k+1}(p)+B_{0}y_{k}(p) \] \[ y_{k+1}(p+1)=Cx_{k+1}(p)+Du_{k+1}(p)+D_{0}y_{k}(p) \] where \(k\geq0\), \(0\leq p\leq\alpha\), \(\alpha\in\mathbb{N}\) is the fixed pass length, \(x_{k}(p)\in\mathbb{R}^{n}\) is the state, \(u_{k}(p)\in\mathbb{R}^{l}\) is the input and \(y_{k}(p)\in\mathbb{R}^{m}\) is the pass profile. This class differs from other classes of 2D linear systems since the information propagation in one of the two independent directions occurs only over a finite duration \(\alpha\). This paper is devoted to the LMI study of the stability along the pass of the considered class of systems. The LMI formulation provides some sufficient conditions for stability along the pass. This approach enables the authors to obtain a stabilizing feedback-feedforward control law of the form \[ u_{k+1}(p)=K_{1}x_{k+1}(p)+K_{2}y_{k}(p). \] This control law has clear physical meaning for practical applications of discrete linear repetitive processes, for instance in iterative learning control. A Lyapunov equation interpretation of stability along the pass is derived and it is used to obtain computable information on the rate of convergence of the output sequence of pass profiles \(\{y_{k}(p)\}_{k\geq1}\) produced by a process that is stable along the pass. An example illustrates the proposed procedures and results. The role of this LMI analysis in the stability and control of repetitive processes in the presence of uncertainty in the model structure is discussed.