Robust Adaptive Control using Multiple-Models, Switching and Tuning

GIOVANINI L, GRIMBLE M

Koss, Grecia

Conferencia; European Control Conference; 2007

European Control Council

An universally accepted solution to the adaptive control problem has been an elusive search. In spite of several decades of research an accepted design methodology for reliable adaptive control, based on sound theoretical issues and suitable for real implementation, has not been found yet. A common problem in many adaptive control approaches is the instability originated by the presence of uncertainties that arise from different causes:* - Interactions between the identification and controller design processes: when adaptation and plant dynamicstime scales are similar, the closed-loop system behavior can result in chaos and instability. Abrupt controller changes: the modification of the controller can introduce new behavior in the overall system that is not present in any of the composite subsystems.* - Poor estimates of parameters: the presence of non-measurable  disturbances and noises lead to poor estimates of parameters and  consequently system instabilities.It is clear that there is a need for adaptive control algorithms with the attributes of non-adaptive robust feedback systems: robust stability and robust performance.