The problem of indirect decentralized adaptive control of a very general class of interconnected systems is considered. The local adaptive controller, which consists of a constrained gradient parameter estimator and a pole assignment control law synthesis module, for each isolated subsystem is designed by ignoring interactions between subsystems and fast parasitics of the subsystem. The a priori knowledge required for the implementation of local adaptive controllers is the (large) range of unknown parameters in the reduced-order models of local subsystems. It is shown that the resulting adaptive system is globally bounded input bounded state stable provided the plant to be controlled has sufficiently weak interactions between subsystems and small unmodeled dynamics of subsystems. No complicated estimator modifications are required to ensure global stability.
展开▼