Recent advances in aircraft technology have led to the development of many new concepts in aircraft design which are strikingly different from their predecessors. The differences are in both aircraft configuration and control paradigms. Considering the vital role of controllers in this area, and difficulties of their implementation on the actual systems, laboratory models like TRMS (laboratory helicopter model) has been manufactured on which the researchers can apply the designed controllers, without doing any harm to the actual systems. However one should first test the controllers using computer simulations, before applying them on the laboratory model. In this thesis modeling of a twin rotor MIMO system is done using ANFIS. ANFIS (Adaptive Network-based Fuzzy Inference System) is a particular type of neuro-fuzzy architectures, which combinates of neural networks and fuzzy inference systems. By taking advantage of its fuzzy rules and hybrid learning algorithm, ANFIS can be used for modeling and control of ill-defined and uncertain systems, without needing complete accurate information about them. Because of these facts, ANFIS has been the center of attention for many of control engineers within the last 15 years. ANFIS model is based on the input-output data pairs of the system under consideration. To improve the performance of ANFIS model in this thesis, subtractive clustering of training data is used to obtain the required initial fuzzy model, and a backward selection method is used to eliminate the unimportant or redundant input variables of the model. Having the initial fuzzy model be extracted with this method, and using only important selected input variables, the final ANFIS model is made simpler and also more accurate. After completing the TRMS model, control of the system is discussed. In this direction, a fuzzy controller which is able to reduce the coupling effects in the twin rotor system, as well as to control each degree of freedom in a reasonable manner, is presented. Althogh this fuzzy controller has good performance, one may encounter problems trying to apply it to the actual plant, due to the high number of rules and high computational time required. In the last section of this thesis, a method for tuning and simplifying the proposed fuzzy controller, using ANFIS, is presented. For this purpose, first an ANFIS structure is designed, second the required training data are extracted, and last the offline training process is done using hybrid learning algorithm. By far the tuned controller is simpler than the initial fuzzy controller; furthermore, offering better performance. The final controller is tested via computer simulations in the presence of disturba Keywords: Twin Rotor MIMO System, Fuzzy logic, ANFIS, MIMO nonlinear modeling, Fuzzy control
