In this article, an iterative learning algorithm based on extended state observer (ESO) is proposed to deal with the propeller failure of an underwater vehicle. In this control scheme, the nonlinear feedback mechanism of ESO is transplanted to iterative learning processes; that is, the nonlinear function of the current output residual is used to adjust the value of the virtual fault in the next iteration. Additionally, to ensure the safety of the control torque, a saturated proportional-derivative (PD) controller is proposed. Finally, to achieve online parameter self-tuning, a fuzzy logic controller is employed in this control scheme to fuzzify the parameters of a saturated PD controller and ESO. The obtained results show the favorable speed of tracking convergence and the high precision of fault estimation.

To deal with thrusters' faults of autonomous underwater vehicle (AUV), an iterative learning algorithm fault-tolerant control (FTC) based on the linear extended states observer (LESO) is proposed. In this control scheme, the non-linear feedback mechanism of the LESO is transplanted into iterative learning processes to estimate fault. Compared to our previous work, LESO is used to substitute classic non-linear extended state observer to make the establishment of the whole system more structured; moreover, the number of parameters need to be tuned can be reduced by the conception of observer bandwidth of LESO. To enhance the controllability and robustness of whole scheme, a new saturated sliding mode controller is proposed based on the Lyapunov theory. Then to achieve online parameter self-tuning for the control system, fuzzy logic controllers are introduced to find optimal relationship between LESO's parameter and tracking errors. The performance of the proposed controller is tested by some comparison experiments on Zhuhai A18D AUV; the results show that the proposed control scheme can ensure better stability than classical control and our previous control scheme when AUV suffers faults.