Fault reconstruction of thruster for autonomous underwater vehicle based on terminal sliding mode observer

• An AUV model based on RBFNN is described in the form of affine nonlinear system.
• A fault reconstruction method based on Terminal sliding mode observer is proposed.
• The effectiveness of the proposed method is demonstrated with pool experiments.

The motion modeling and thruster fault reconstruction for autonomous underwater vehicle (AUV) system are addressed in this paper. Considering the modeling uncertainty of the AUV motion model given by dynamics analysis method, we present an AUV motion modeling method based on RBF neural networks. Since there is asymptotic convergence problem in the process of using traditional sliding mode observer to estimate the state signal, which cannot be measured directly by sensor, the fault signal cannot be reconstructed timely. Therefore, a Terminal sliding mode observer is presented to ensure each estimated state signal converge in a finite time. According to the output of Terminal sliding mode observer, the equivalent output injection method is used to reconstruct thruster fault. Finally, the feasibility and effectiveness of the proposed approach is demonstrated with pool experiments of the experimental prototype.