FUZZY CMAC NEURAL NETWORKS AND FLIGHT VEHICLE CONTROL

Future flight vehicles will require incredased speed, precision, and survivability under a wider variety of flight conditions than current technology supports. The traditional role of flight control systems (FCS) in providing stability and command augmentations must be expanded to achieve enhanced maneuverability, combat effectiveness, and self repair. Recent advances in the development of fuzzy logic controllers and neural networks provide a unique opportunity for achieving highly integrated, reliable, robust and adaptive fllight control systems which enhance maneuverability, accommodate uncertainties,and simplify design and maintenance. During the proposed work an innovative fuzzy control system architecture, namely the Fuzzy-CMAC (cerebellar model arithmetic computer) neural network, will be evaluated for use in advanced flight control systems. This architecture exploits a synergism between the CMAC neural network and fuzzy logic control. CMACS and fuzzy logic are similar in terms of input encoding and mapping structure. Different features of both models are compared and a Fuzzy CMAC model which combines preferred features of both models is then established. The fuzzy CMAC controller utilizes available human control experience in the initial construction of its knowledge base. Then the CMAC neural network further provides powerful self-learning capability and flexibility in achieving robust performance over an expanded fligh envelope.