Abstract: State observation is a popular model analyzing method in fault diagnosis and fault-tolerant control, but its application in uncertain nonlinear systems is restricted because of the existence of unknown input disturbance. In order to increase both the robustness for model uncertainties and the tracking capacity for actuator faults in the uncertain nonlinear electromechanical systems with type of strict feedback, an integrated fault diagnosis and fault-tolerant control program is proposed in which the unknown input disturbance is considered. A fault-tolerant controller is designed on the basis of sliding mode variable structure theory, and a design of state observation follows by use of a specific equivalent control methodology in this theory. Various types of faults are reconstructed by means of the adaptive method. Simulation analysis of this program in electro-hydraulic servo systems shows that it is robust to model uncertainties and relatively more able to reconstruct not only abrupt, intermittent variations such three common faults, but also the noisy slow-variation faults.

Key words: Electro-hydraulic servo system, Fault reconstruction, Fault-tolerant control, Observers, Sliding mode