Didier Theilliol

Actuator fault detection, isolation method and state estimator design for hot rolling mill monitoring

This paper proposes a method for actuator fault diagnosis in a speed control loop of a hot roughing rolling mill used to carry out multi-pass forward-forward or forward-reverse rolling. The main objectives of the work is to first improve an actuator fault detection and fault magnitude estimation problem for a particular system and second to estimate non measurable variables in order to provide suitable information to the monitoring system. According to fault diagnosis theory and the specific application domain, a model-based fault detection filter combined with a conventional Luenberger observer is proposed. Based on an appropriate gain, the developed technique enables to supervise the hot roughing rolling mill. The effectiveness and performances of the technique are illustrated on a computer simulation platform.

Active Actuator Fault Tolerant Control Design for Polytopic LPV Systems

In this paper, an Active Fault Tolerant Control (FTC) strategy is developed for polytopic Linear Parameter Varying (LPV) systems. The main contribution concerns the design of a linear Output Feedback dedicated to polytopic LPV systems. Based on an available Fault Detection, Isolation (FDI) scheme, the controllers redesign is performed on-line trough LMI both in fault-free and faulty cases in order to preserve the system closed-loop stability despite of actuator failures. The effectiveness and performances of the proposed FTC method are illustrated in a simulation example.