A new unmatched-disturbances compensation and fault-tolerant control for partially known nonlinear singular systems

This study investigates the challenge of designing a fault-tolerant control (FTC) for partially known nonlinear singular systems subjected to unmatched uncertainties in addition to actuator and sensor faults. The suggested technique is dependent on the neural network-based adaptive observer. The unknown system nonlinearities are approximated by making use of an adaptive neural network (NN) approximation technique. The parameters of the NN are unknown. A new methodology is proposed to transform the partially known singular system to a non-singular form with unknown uncertainties. Different from all previous work dealing with singular systems with partially known system parameters, an adaptive observer is proposed with the help of adaptive laws to obtain an estimation of the augmented states of the constructed descriptor system. Based on the estimated states, a new approach for dealing with unmatched disturbances and faults are proposed. Finally, an adaptive controller is designed to account for unmatched disturbances and faults. The asymptotic stability of the overall system is guaranteed via Lyapunov-stability functional. The designed method is applied efficiently to a satellite control system as a practical example in addition to another numerical example.