scientific article; zbMATH DE number 1474025

From MaRDI portal
Publication:4490259





This booklet introduces to the control theory of finite-dimensional systems in Euclidean space by focusing on the fundamental areas of controllability, stabilizabiliy and optimal control. The introduction first motivates control theory, then studies the well-posedness of some perturbed linear time-varying ODEs, introduces the fundamental matrix and its properties and finally states several fixed point theorems with a proof of the Banach fixed point theorem. Chapter two is devoted to observability. In the linear case, the nonsingularity of the observability Grammian allows to recover several criteria including the well-known Kalman observability criterion. The reconstruction kernel is introduced. In the nonlinear case, one splits the right-hand side of the differential equation into a time-varying linear part and a nonlinear perturbation. The authors derive an expression for the state as an integral equation (depending on the inverse of the observability Grammian, the system data and the fundamental solution to the drift) with two contributions: one from the output and one from the nonlinear part. This allows to establish a global observability result under two conditions: the determinant of the observability Grammian is strictly positive and the nonlinear perturbation is small enough so that a fixed point theorem can be applied. Chapter three is devoted to controllability. In the linear case, one introduces the controllability Grammian and one proves a duality result between controllability and observability. This allows to obtain similar properties as in chapter two by duality, but the Kalman controllability criterion is proved directly. Next, the steering function is introduced and allows to obtain analytical formulas in several examples. The nonlinear case is handled by adding a nonlinear function (of the state and the control) to the right-hand side of the differential equation; it is assumed that it does not grow too much asymptotically (in particular it is sublinear asymptotically). Then, the authors prove a complete controllability result using the Schauder fixed point theorem. One uses a mapping between (state, control) pairs in the Banach space of continuous functions on an interval of the line. The new control is such that it uses the steering function for the linear part and compensates for the perturbation dependent on the old control. The terms eliminate in the equation for the state when the new and old control coincide. The fixed point theorem can be applied because of a result of the second author about the behaviour of the nonlinear term. A further section establishes complete controllability results in similar situations when moreover the values of the control are prescribed at the end points of the interval (or in between). A last section deals with asymptotic null controllability (ANC). The controllability Grammian is assumed to have a limit when time becomes infinite and the nonsingularity of the limit is equivalent to ANC in the linear case. The authors use the Lévy-Schauder theorem in the nonlinear case under suitable assumptions to get ANC. Chapter four presents several results on the stability of nonautonomous possibly nonlinearly perturbed linear systems. Some of them depend on assumed properties of the fundamental matrix of the linearization. Chapter five is on the stabilizability of linear autonomous control systems. Stabilizing controllers are obtained. They depend on a controllability Grammian or on the solution to a Lyapunov equation. The Kalman decomposition and output stabilization issues are included. The last chapter handles optimal control of linear control systems (the LQ theory) first in a time-varying context (finite horizon) and then in the autonomous case (infinite horizon). The extremum principle is used and the solution is seen to depend on the solution to the Riccati equation. In the second case the solution is seen to be stabilizing. A case with a nonlinear perturbation where one uses again a fixed point theorem ends this part. A bibliography with 25 entries (most of them books and many of them from the sixties and seventies the most recent one from 1991) and a small index end the volume. This pleasantly written book is concise and includes many examples which illustrate the themes treated. This introductory text presents classical topics, following the ``trinity of control theory: [(observability, controllability), stabilization, optimal control]. They focus on finite-dimensional, linear possibly nonlinearly perturbed systems. This allows to orient the presentation through the frequent use of the fundamental matrix of the linearization and some fixed point theorems, which permit to obtain nonconstructive existence and uniqueness results. These methods are a speciality of the authors who used them in many of their publications in particular for delay, neutral or (and) integrodifferential systems in abstract spaces. This good book can be recommended as an introduction to control theory for students or for persons who want to know fundamentals of this area of contemporary science. They will be quickly led to what is important.











This page was built for publication:

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4490259)