A DC motor fault detection, isolation and identification based on a new architecture Artificial Neural Network

Any diagnosis procedure should be rapid and efficiency. In the presence of the noise and unknown inputs, the diagnosis procedure could generate false alarms. This paper focuses in the fault detection, isolation and identification when a fault affected the sensors of a DC motor worked in disturbance environment. The chosen technique should take account the two issues: the noise and the unknown inputs. As a solution the Artificial Neural Network is adopted in order to generate a robust residuals and therefore to minimize the false alarms. A new ANN architecture is proposed to achieve this purpose. The simulation results proved the rapidity and the efficiency of the proposed procedure.