Improved Fault Detection for Nonlinear Mechatronic System based on Uncertain Bond Graph

This paper deals with the improved fault detection for nonlinear mechatronic system based on the uncertain bond graph. Parameter uncertainties are taken into consideration and are represented in linear fractional transformation (LET) for calculating the improved adaptive threshold. The uncertainties are incorporated in the calculation of the analytical redundancy relations whose numerical evaluations arc used for fault detection purpose. The main contribution of this work aims to eliminate uncertainties from the traditional adaptive threshold and obtain the improved one which is less conservative. Therefore, the sensitiveness of fault detection has been improved by utilizing the improved adaptive threshold. The practical implementation of the method is validated by a mechatronic system with a motor reducer and an external load. The simulation results shows the effectiveness of the proposed method.