Abstract: The mechanism that the control surface failure affects the movement of aircraft is investigated,and the movement modelling under control surface failure is systemically expounded.The reconfigurability of the faulted system is researched based on modelling.Two methods are introduced to assess the level of reconfigurability.One is deduced from linear system theory,and the other is defined from the opinion of physical movement.The learning algorithm of propagation forward neural network which is adopted in control system design is proved.In this propagation forward neural network,the error is back propagating.A novel architecture of reconfigurable flight control that introduces back propagation neural network to compensate the normal controller is proposed.Simulation on the controller with nonlinear aircraft movement model shows the reconfigurable control system operates well.