Abstract: A robust adaptive reconfigurable flight control method is presented based on fuzzy model. Based on T-S fuzzy model, the overall control scheme is constructed by combining all local linear regulators and robust adaptive neural network reconfigurable controllers. The global asymptotic stability of closed-loop reconfigurable system is proved. It provides the solution to overcome the drawback of the conventional gain scheduling that the parameter change may be rather abrupt across the boundaries of the region, and the control law does not globally posses the desirable properties designed at local operating point. The learning ability of proposed full adaptive RBF neural network is improved through adaptive tuning of the weights, centers and widths on line. And the compensate component of RBF neural network is designed to on-line estimate the bound on the neural network approximate error. It can be used to cancel the effect caused by the modeling error, disturbance or system failure, and maintain the handling qualities of the system. Simulation shows that the reconfigurable method receives a good effect.

Key words: T-S fuzzy model, reconfigurable control, adaptive neural network, fighter, robustness