Sensor fault diagnosis of inland navigation system using physical model and pattern recognition approach

In recent years, inland navigation networks benefit from the innovation of the instrumentation and SCADA systems. These data acquisition and control systems lead to a reactive asset-management of inland navigation networks. However, sensors and actuators are subject to faults due to the strong effects of the environment, aging, etc. In this paper, a sensor Fault Detection and Isolation (FDI) approach is proposed using an Integrator-Delay-Zero (IDZ) model, interval observers and the dynamic classification algorithm AUDyC. The combined use of these approaches allows the improvement of the sensor fault diagnosis. The proposed approach is introduced through the case study of the Cuinchy-Fontinettes reach in the north of France.