Abstract
The work presented in the thesis concerns the design and development of a novel thruster fault diagnosis and accommodation system (PDAS) for overactuated, open-frame underwater vehicles. The remotely operated vehicles (ROVs) considered in this thesis have four thrusters for motion in the horizontal plane with three controllable degrees of freedom (DoF). Due to the redundancy resulting from this configuration, for the case of a partial fault or a total fault in a single thruster it is possible to reallocate control among operable thrusters in order that the ROV pilot is able to maintain control of the faulty ROV and to continue with missions.The proposed PDAS consists of two subsystems: a fault diagnosis subsystem (FDS) and a fault accommodation subsystem (FAS). The FDS uses fault detector units to monitor thruster states. Robust and reliable interrogation of thruster states, and subsequent identification of faults, is accomplished using methods based on the integration of self-organising maps and fuzzy logic clustering. The FAS uses information provided by the FDS to perform an appropriate redistribution of thruster demands in order to accommodate faults. The FAS uses a hybrid approach for control allocation, which integrates the pseudoinverse method and the fixed-point iterations method. A control energy cost function is used as the optimisation criteria. In fault-free and faulty cases the FAS finds the optimal solution, which minimises this criteria. The concept of feasible region is developed in order to visualise thruster velocity saturation bounds. The PDAS provides a dynamic update of saturation bounds using a complex three-dimensional visualisation of the feasible region (attainable command set), such that the ROV pilot is informed with the effects of thruster fault accommodation, incorporated in the new shape of the attainable command set. In this way the ROV pilot can easy adapt to newly created changes and continue the mission in the presence of a fault.
The prototype of the PDAS was developed in the MATLAB environment as a Simulink model, which includes a nonlinear model of an ROV with 6 DOF, propulsion system and a hand control unit. The hand control unit was simulated in hardware using a joystick as input device to generate command signals. Different fault conditions are simulated in order to investigate the performance of the PDAS. A virtual underwater world was developed, which enabled tuning, testing and evaluation of the PDAS using simulations of two underwater vehicles (FALCON, Seaeye Marine Ltd. and URIS, University of Girona) in a 'realistic' underwater environment.
The performance of the PDAS was demonstrated and evaluated via tank trials of the FALCON ROV in QinetiQ Ocean Basin Tank at Haslar, where the existing control software was enhanced with the PDAS algorithm. The results of real-world experiments confirmed the effectiveness of the PDAS in maintaining vehicle manoeuvrability and in preserving the vehicle mission in the presence of thruster faults.
Date of Award | Feb 2004 |
---|---|
Original language | English |
Keywords
- Fault diagnosis
- Fuzzy logic
- Engineering
- Mechatronics
- Remotely operated underwater vehicles (ROV)
- overactuated open-frame underwater vehicles
- Fault Diagnosis and Accommodation System (FDAS)
- control systems