scientific article; zbMATH DE number 3914194

From MaRDI portal
Publication:3690692





The survey reflects recent advances in some selected topics of distributed parameter system (dps) control theory. These fields are: (1) structural dps theory, (2) mobile dps control, (3) control of quantum processes, (4) differential geometric methods applied to the problem of controllability and existence of finite controls. The first trend stems from the desire to provide a software able to model processes in complex distributed systems, i.e. those assembled of the elementary blocks described by functional equations of various types. The combination of such blocks establishes a unique (analytical or algorithmic) relation between the distributed input and output signals. The software consists of three kinds of programs: (i) A library (file) of programs for computer implementation of characteristics (commonly Green functions) of elementary distributed blocks. (ii) Programs for computing typical combinations of two elementary distributed units (sequential connection, parallel connection, feedback loop). (iii) Programs for transforming structural diagrams of complex distributed systems. These programs describe the interrelationship between signals and units of the structural diagram. The second trend concerns a special class of controls depending on the argument x-s(t) where \(x\in R^ n\) denotes coordinates and t denotes time. This type of control often arises in problems of control heating processes, e.g. stabilization of temperature distribution etc. Several examples are discussed. The third trend is concerned with a very interesting class of control processes which, however, cannot be associated with dps, since the control objects are governed by ordinary differential equations. The equations are those of quantum mechanics where the potential of the external field is treated as a control. A number of optimization problems are stated, e.g. those of maximization of probability of the desired value of a given physical quantity, of the best approximation of some given distribution of eigenvalues of the physical quantities etc. Some examples of controlled quantum processes are referred to. The last topic is associated with the problem of qualitative description of the field of extremals for control problems stated for ordinary differential equations. The suggested construction is based on the determination of the so-called boundary of the integral funnel P(q') for the contingency equation which is equivalent to the initially given control system \(\dot q=f(q(t),u(t)),\) \(q(t)\in R^ n,\) \(u(t)\in U,\) \(t>0.\) The contingency equation is of the form \(\dot q\in f(q(t),U)\) where f(q,U) denotes the convex closure of \(\{\) f(q,n)\(q\in {\mathbb{R}}^ n\), \(u\in U\}\). The surface P(q') is obtained by integration of the Hamilton-Jacobi equation associated with the stated augmented control problem using the Pontryagin maximum principle. Specifically, this boundary is the surface of the characteristic conoid of the Hamilton- Jacobi equation. The surface P(q') is supplied by hatching, i.e. it is hatched at that side from where there may come an admissible path. The field of hatched surfaces P(q') provides in this way a topological description of the field of extremals and proves useful in solving the problem of controllability, existence of sliding regimes, etc.











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 Q3690692)