Jérôme Cieslak

A LPV approach for early fault detection in aircraft control surfaces servo-loops

Abstract The work presented in this paper is undertaken within the European FP7 funded ADDSAFE * project. The goal is to propose new fault diagnosis techniques that could significantly help developing environmentally-friendlier aircraft, by optimizing structural load design objectives. In this paper, a Linear Parameter Varying (LPV) model-based fault detection scheme is proposed for robust and early detection of faults in aircraft control surfaces servo-loop. The proposed methodology is based on the H ∞ / H _ LPV technique proposed in (Henry, 2012), for LPV systems modeled in a linear fractional representation (LFR) fashion. Simulation results using a “high-fidelity” simulator provided by AIRBUS, are presented to demonstrate the potential of the proposed technique.

A Method for Actuator Lock-in-place Failure Detection in Aircraft Control Surface Servo-loops

This paper deals with a signal-based method for robust and early detection of lock-in-place failures (a.k.a. jamming) in aircraft control surface servo-loops. Earlier and robust detection of such failures is an important issue since they may cause additional structural load and affect the sustainability of civil transport airplane. The proposed signal-based scheme uses a sliding-mode differentiator to provide derivatives of measurable signals in noisy environment. Jamming events are next detected by using a dedicated decision making-rule that is able to detect actuator outage (the stuck value can be near zero). The proposed monitoring scheme has been tested on Airbus test facilities located at Toulouse, France. The results confirm good level of robustness and performance, even in extreme situations. The proposed technique can be applied, with slight modifications, to any type of actuator, e.g. Hydraulic, Electro-Hydrostatic (EHA) or Electro-Backup-Hydrostatic (EBHA) actuators.

Supervisory Fault Tolerant Control Scheme based on Bumpless scheme and Dwell-time Conditions

Abstract The paper deals with the design of an active Fault Tolerant Control (FTC) strategy based on supervisory control approach for Linear Time Invariant (LTI), disturbed and uncertain systems. From a bank of pre-defined Luenberger observers, the observer that yields the smallest observation error determines the present operating mode that is used by a supervisory algorithm to select the corresponding controller. The cases of intermittent and abrupt faults are covered. A key feature of the proposed approach is the establishment of a set of conditions providing transient management and improved value of dwell-time for fault tolerant controllers possessing an integral action or unstable poles. The efficiency of the approach is demonstrated on the HiMAT benchmark (Hou et al., 2010). Copyright

Enhanced distinguishability in Supervisory Fault Tolerant Control

This paper deals with the design of an estimator-based supervisory Fault Tolerant Control (FTC) scheme for Linear Time Invariant (LTI) systems. A formal stability proof based on dwell-time conditions is established in case where the state of the detection filter does not correspond to the plant state, as in classical (Luenberger) observer-based approaches. In this context, fault isolability could be improved leading to an enhanced distinguishability between all possible operating modes. The efficiency of the proposed technique is illustrated on a numerical example.