Nonlinear fault tolerant control of nonlinear structure with MR dampers

For civil structure with control system, damages to structure or damages to sensors and actuators of the control system may occur especially during an intensive earthquake. Therefore, fault tolerant control is becoming more and more important in control engineering and civil engineering. In this paper, an innovative dynamic neural network based fault tolerant control is proposed for nonlinear structure considering actuator faults. By means of the approximation ability of the dynamic neural network, a well designed dynamic neural network model is used to identify the nonlinear structure with actuator faults. The training algorithm of this neural network is derived from Lyapunov direct theory. Based on this dynamic neural network model, the corresponding fault tolerant control is designed and analyzed. The simulation on the nonlinear benchmark structure is conducted and studied. In order to contrast with other controllers, a previously designed intelligent controller is also simulated to verify the effectiveness of the proposed fault tolerant controller. Simulation results show that the proposed dynamic neural network based identification algorithm can satisfyingly approximate the nonlinear structure with different actuator faults. Moreover, the further developed fault tolerant controller is able to maintain stability and acceptable degree of control performance not only when the structure is fault-free but also where there are actuator faults.