Adaptive control algorithms with multiple models and fuzzy logic switching

Three different designs of multiple model adaptive controllers using fuzzy logic switching have been developed in this dissertation to cope with complex and difficult environment conditions that the plant under control might undergo. The main concept of each proposed design is to ensure such automatic change of the controller parameters so that they correspond to the current plant environment and provide an appropriate control action to improve the overall control system performance.; The three different designs have been developed in the contexts of MRAC and STR systems. In MRAC system, a switching block at each instant receives some information from the command signal and any auxiliary output as a criterion to be used in the fuzzy rules as inputs to select the suitable reference model for that instant. In STR system, two designs are developed. The first design is parallel controllers. In this design a switching block using fuzzy rules performs the logical and functional paralleling of a conventional fixed parameters controller and an adaptive controller. The switching block receives the identification error and its derivative as a criterion to evaluate the weighting factor that combines the two controllers to derive the total control signal that applied to the plant. The second design is using multiple identification models working in parallel, at each instant the switching block evaluates some information obtained from auxiliary outputs as a criterion to select the most suitable identification model for that instant.; The effectiveness of the proposed techniques is demonstrated on the applications of power system, induction motor and robotic manipulators using computer simulation studies. The obtained results show that an improvement in the overall system performance is obtained using the proposed multiple model adaptive controllers in comparison to those of single model adaptive controllers as well as conventional fixed parameters controllers.