Sensor fault detection, isolation and recovery design applied to a winding machine
Résumé
The paper presents the results of a complete study on sensor fault detection, isolation and recovery (FDIR) applied to a winding machine. The process is modelled by a Linear Time Varying (LTV) model. The control of web tension is based on estimated or virtual measurements as feedback variables. The main contribution concerns the design of an accurate bank of observers able to isolate and estimate sensor faults while preserving the system's integrity in terms of structure and performance. The objective of FDIR, which is also called Active Fault Tolerant Control System, is to maintain current performances close to desirable ones and preserve stability conditions in the presence of faults. The paper addresses a general method that could allow to detect, to isolate, and to estimate sensors faults. An FDIR for LTV system is developed and preserves sensor fault through appropriate fault detection, isolation module. The effectiveness and performances of the strategy are illustrated on a winding machine in the case of a complete loss of sensor.