In the last years, due to the rapid developments of the technology, the use of underwater vehicles has remarkably increased as a cost-effective solution for performing tasks such as subsea observations, the monitoring and maintenance of off-shore structures or pipelines and the exploration of the sea bottom. the possible occurrence of fault in underwater vehicles is one of the most common problems in control and navigation of these vehicles. The occurrence of fault may greatly affect the control performance of the underwater vehicles leading to the decreasing of working efficiency and increasing of working cost. Therefore, it is crucial to develop a fault tolerant control system (FTCs) inorder to ensure the reliability and capability of performing missions. In this thesis, the problem of fault detection, estimation and design of fault tolerant control system for an underwater robot has been considered. The vehicle is equipped with four thrusters, controlling its position and orientation in planes parallel to the sea surface, and is connected with the surface vessel by a supporting cable. This vehicle model is nonlinear. It is assumed that one of the four thrusters can undergo a fault. The fault is modeled as actuator loss of efficiency or total actuator failure. First, a method is used to detect and estimate the fault in the underwater robot. For this purpose, a fault diagnosis scheme based on two Thau observers is considered to detect and estimate the fault severity. In this regard, the Thau observer is used for system monitoring and residual generation. Then, an adaptive Thau observer is activated to estimate the magnitude and severity of the fault. In the following, the scheme of fault tolerant control which composed of fault detection and estimation module, nominal controller based on sliding mode control and a mechanism to compensate the fault effects is expressed. Simulation results illustrate the performance of the scheme. Key words Underwater vehicles, fault detection and estimation, Thau observer, adaptive observer, fault tolerant control.
