This paper introduces a novel fault diagnosis system of thruster for open-frame underwater vehicles (UV). Basically, the fault diagnosis system is a control allocator, but this primary function is enhanced with the ability of automatic thruster fault detection and diagnosis. The proposed fault diagnosis system consists of two subsystems: a underwater vehicle motion condition monitoring subsystem and a thruster fault detection subsystem. The motion condition monitoring subsystem analyses the output residuals of UVs motion model which is constructed with an improved Elman neural network and the real state value, to monitor its motion condition. The thruster fault detection subsystem uses fault detector units associated with each thruster, to monitor their conditions. Robust and reliable fault detection units are based on RBF neural network and fault conditions classifying methods. Fusion fault diagnosis unit integrates information provided by the two subsystems to locate and identify thruster fault. Results of the actual experiment show that the proposed fault diagnosis system is effective and feasible
[1]
Meyer Nahon,et al.
Dynamics modeling and performance evaluation of an autonomous underwater vehicle
,
2004
.
[2]
Lingli Ni,et al.
Fault-Tolerant Control of Unmanned Underwater Vehicles
,
2001
.
[3]
G. N. Roberts,et al.
Thruster fault diagnosis and accommodation for open-frame underwater vehicles
,
2004
.
[4]
Tamaki Ura,et al.
Development of a system to diagnose autonomous underwater vehicles
,
1999,
Int. J. Syst. Sci..
[5]
Simon Haykin,et al.
Neural Networks: A Comprehensive Foundation
,
1998
.