Skip to Main Content
Adaptive control using multiple identification models is currently well established. Switching, tuning, and switching and tuning in such systems have been extensively investigated, and their stability properties have been established . If the number of models is sufficiently large, and is uniformly distributed in parameter space, the method results in fast and accurate response. In practice, even for reasonably complex dynamical systems, the number of models needed for improved performance may be quite large. The paper attempts to propose a general methodology for achieving comparable response using significantly smaller number of models. The principal idea is to either redistribute available fixed and adaptive models even as the system is in operation, or utilize adaptive models more efficiently. Several different approaches are described and in each case the theoretical and practical questions that arise are discussed. Simulation results are included to demonstrate the improvement in performance using the methods proposed.