Executive-controlled adaptive systems

THERE IS an immense body of literature on the general subject of feedback control theory. The greatest portion of this literature is devoted to the analysis and synthesis of single-loop or simple multiloop linear or piecewise-linear feedback systems. In most of this work, there is a common fundamental assumption, and that is that the designer has, or can obtain, reasonably complete knowledge of the properties of the device or process which is to be controlled. This knowledge generally consists of a fairly accurate characterization of the process or device under one set of environmental conditions, along with information indicating the manner and the limits within which the properties of the process vary with time and with different external conditions. It is the ability of closed-loop control to render the complete system relatively insensitive to inaccuracies in the characterization and to variations in the properties of the process to be controlled that has resulted in the phenonomenal growth of the field.