| Performance evaluation is one of the major problems encountered during the design of nonlinear feedback control systems. Typically, designers are concerned with both a system's transient and steady state performances. It is possible to examine the stability of a nonlinear system using Lyapunov's second method, and determine a region within which the system is stable. However, stable system operation in the sense of Lyapunov may not produce satisfactory performance, as the rate of convergence of an asymptotically stable system may be unacceptably long in some regions of operation. The inclusion of plant uncertainty adds to the problem. Although robust performance design techniques can be used to design a controller that minimizes the impact of plant uncertainty, it does not provide any indication of how the uncertainty effects the system's transient behavior in the time domain.; In this thesis, performance criteria are developed for a class of nonlinear systems including plant uncertainty based on the difference between the system's response and the response of a reference linear system. To illustrate the usefulness of the criteria, applicability to the analysis of feedback control systems utilizing inverse dynamics and perturbation theory is investigated. In addition, the applicability to the analysis of neural network based feedback control systems is also studied. Finally, the performance criteria shall be used to study the performance of various feedback control systems. |
