Extended State Observer Based Iteration Learning Fault-tolerant Control Scheme for AUV

Fuzzy linear extended states observer‐based iteration learning fault‐tolerant control for autonomous underwater vehicle trajectory‐tracking system

“…To reduce process of parameter tuning and non‐linear function design, in this work, ESO is substituted by LESO. Like the conception of ESO in [18] the attenuation or extinction of power acting on AUV can be regarded as disturbance from the opposite direction. In this paper, like the mechanism of iterative ESO in our previous work, the non‐linear feedback of the LESO designed from the time axis is extended to iterative axis in order to improve the convergence speed of virtual fault and put forward a kind of iterative learning‐based fault‐tracking estimator.…”

“…Since the fault estimator is proposed by using the current iteration of the output error messages only, the algorithm operation is simple, easy to implement in engineering without using the differential signal of output error. In our past work, to ensure safety of whole control system, a non‐linear saturated PD controller is designed as a key controller [18], the results showed some advantage than the classic controller, while robustness is an unavoidable imperfection of PD. It means that the previous controller is good at addressing single faults, while in latest tests we find that it cannot perform well when dealing with some new kind of thruster faults or internal uncertainties.…”

“…Inspired by the above analysis, in order to address faults/failure of thrusters and disturbance simultaneously, we have used an extended state observer (ESO) to track and estimate the synthesis of faults and external disturbance in our previous work [17,18]. The key thought of ESO is to treat the internal uncertainties and external disturbances as a 'generalized total disturbance', and try to estimate it in real time.…”