Identification and controlling of linear systems: Utilizing adaptive inverse control systems and adaptive hybrid algorithms

The main objective of this dissertation is to identify and control linear plants by employing a new non-conventional algorithm that constructs models topology, and adapts the constructed models parameters. A hybrid system (HS) was employed to construct either a plant model and/or a controller model. The new non-conventional learning algorithm is called an adaptive hybrid algorithm (AHA). This algorithm employs HS to construct the models, and then adapts the parameters of these models. Moreover, the inverse techniques of minimum or nonminimum phase single-input single-output (SISO) linear systems were employed also by AHA to construct the control of these systems.; Basically, this dissertation is concerned with the following tasks: (1) Identifying a given IIR plant with an IIR plant model. (2) Constructing the inverse controller of a given IIR plant with two different system models. (3) A ringing phenomenon that is inherent in inverse control schemes was lessened by combining a Pole Shifting Compensation Process (PSCP) with AHA. However, all of the preceding tasks were accomplished using adaptive hybrid algorithm AHA. (4) Infinite Impulse Response Least Mean Square (IIR-LMS) algorithm suffers from slow convergence rate and destabilization, but choosing the appropriate step size for each parameter of the IIR model in each iteration would improve the IIR-LMS algorithm. Therefore, IIR-LMS was modified with fuzzy learning rate where the modified IIR-LMS is called FIIR-LMS.; Furthermore, it would be shown that AHA is characterized with two stages. Stage one of AHA is a combination of Infinite Impulse Response and Genetic Algorithm (IIR-GA), and it is employed by AHA to predict the structure or topology of IIR plant model and/or IIR controller model of the given IIR plant. Stage two of AHA is a combination of Genetic Algorithm and Fuzzy Infinite Impulse Response Least Mean Square (GA-FIIRLMS) and it is utilized by AHA to adapt the constructed hybrid topology weights or parameters of the model.; In addition, simulation results for the aforementioned scheme are presented to confirm the efficacy of the proposed method.