The Identification of a Class of Nonlinear Control Systems by Elimination Regression

摘要

Work in nonlinear control system identification has proceeded slowly because a general technique for solving nonlinear differential equations is not known. Many techniques have been developed in nonlinear identification, but most are oriented toward small classes of problems rather than toward the large class of all nonlinear systems with lumped elements. A motivation for the work has been to develop one of these techniques further than it has been developed and to do so preferably in a manner that will create an identification scheme that is simple to apply. Identification by regression methods was the technique chosen for investigation. Elimination regression can be applied in the identification of man-made and biological control systems, in modeling problems, and in supplementing other identification techniques. Elimination regression is illustrated in an application to a computer-generated problem involving a second-order system. The potential for extension to systems other than those described in the problem statement is demonstrated in a biological modeling problem. The latter problem is a part of the larger problem of modeling the relation between heart rate and respiration.