Optimal and minimum-energy optimal tracking of discrete linear time-varying systems (Q1902570)

The authors consider discrete time linear time-varying systems. For these systems solutions for the optimal minimum energy tracking problem with finite time horizon are developed, where arbitrary time-varying weight matrices are permitted. The main mathematical tool used is a generalization of the Moore-Penrose inverse by observing that the requirements related to optimality and minimum energy can be separated. After discussing these facts two algorithms are presented for the optimal tracking problem: a direct one followed by a numerically more efficient one in the sense that the size of the matrices that have to be computed does not depend on the time. The special case in which the tracked trajectory equals zero is discussed separately. In the next step these algorithms are extended to the minimum energy case. In addition, explicit expressions for the tracking error and the minimum energy are given. Apart from the facts stated above the proofs are based on Bellman's optimality principle and extensive matrix manipulations. Finally, solutions for the special cases of the terminal control problem for time-varying as well as for time-invariant systems are given, followed by some remarks on reachability and controllability and on the dependence of the minimum energy on the terminal time.