Design of a Computer Algorithm for Fine- Tuning of Adaptive Controllers

In this thesis a direct adaptive model-based control algorithm is designed.Most existing methods of adaptive control employ some kind of plant modelwhich is used to infer the error of the control signal from the error at the plantoutput. This error is used to adjust the controller parameters -such that somecost function is optimized. Schemes of this kind are generally described asbeing indirect.Unlike these, the designed algorithm is direct SInce it calculates thecontrol signal error by performing input matching. This requires generatingtwo control signals; the first control signal is applied to the plant and thesecond is inferred from the plant response. The controller error is thedifference between these two control signals and is used by the algorithm toadapt the controller.The method is shown to be a viable strategy for adaptation of controllersbased on inverse process model, adapted via input matching techniqueperforming non linear function approximation. It is proven that: provided thata given controller is sufficiently close to optimal at the commencement ofadaptation process, its parameters will converge, and the control signal andthe output of the plant being controlled will be both bounded and convergent.To demonstrate the applicability of the algorithm; a neural controller wasimplemented, and computer simulations were performed for some non-linearprocesses. Simulation Results demonstrated that the algorithm yields asystem performance that is comparable or superior to that of other neural andlinear adaptive control paradigms.