Deterministic learning and fault diagnosis for nonlinear robotic manipulators

The diagnosis of faults is one of the important tasks in the operation of robotic manipulators. In this paper, a rapid fault diagnosis scheme is proposed for nonlinear robotic systems. Firstly, the system uncertainty and unknown fault dynamics are identified through deterministic learning. The knowledge on uncertainty and fault dynamics is stored in a bank of neural networks (NNs). Secondly, a mechanism for rapid fault detection and isolation (FDI) is presented, by which a fault occurred can be detected and isolated by smallest residual principle. Simulation studies are included to demonstrate the effectiveness of the proposed approach.