Application of H-infinity fault diagnosis to ADDSAFE benchmark: the control surface jamming case

Abstract

In this article the H  fault detection and diagnosis technique is applied to a representative twin-engine civil commercial aircraft. The specific failure scenario contemplated concerns the detection of an abnormal aircraft behavior leading to the degradation of the aircraft performance. This abnormal configuration is caused by an actuator or a sensor failure in the control loop of a control surface, between the Flight Control Computer and the moving surface, including these two elements. The definition of the aircraft benchmark, model and fault problematic, and the development and application of the FDD technique application is part of a European Framework Programme project named Advanced Fault Diagnosis for Sustainable Flight Guidance and Control. The results, including nonlinear time domain simulations in the closed-loop functional engineering simulator, show the promising performance and robustness of the proposed solution.

Second order sliding mode observers for the ADDSAFE actuator benchmark problem

“…The work described in [31,32,33,34,35,28] are examples of other approaches to actuator jam/runaway failure detection within ADDSAFE. The work in [31,34] uses a mathematical model of the actuator to design the FDD scheme and therefore requires a low order observer scheme.…”

“…In [35] an LPV representation of the nonlinear actuator model is used as the basis for the observer design. The work in [32,33] consider a more 'global' approach where a typical lateral aircraft model is used as the basis for an observer to generate residuals which are evaluated to detect the presence of an actuator jam or runaway.…”

Advanced sensor fault detection and isolation for electro-mechanical flight actuators

“…Advanced methods as LPV techniques in [2] and [3], sliding mode observers in [4], nonlinear observer based approaches in [5] and H ∞ techniques in [6] have been applied to monitor hydraulic actuator systems of modern aircraft locally, i.e., by looking at the input-output dynamics of the actuator. System wide monitoring approaches using the input-output behavior of the overall aircraft dynamics are provided by [7] and [8]. [9] proposes a hybrid system combining local and system wide fault detection approaches.…”

Enhanced fault detection and isolation in modern flight actuators

“…Pure global approaches are provided e.g. by [3] and [4]. While various approaches are able to detect faults in an actuation system quite well, none of them tries to distinguish between physical faults of the actuator and faulty sensors, necessary in the actuation system to close the actuator's control loop.…”

“…additive sensor fault vector f s allows the modeling of three sensor faults f s,1 , f s,2 and f s,3 . For example a freezing of the third sensor at the position c f is created by selecting f s,3 = −x + c f .Setting the parameters ρ in(3) to constant values and including the defined fault and noise inputs leads to the first order linear model of the actuation systeṁ…”