A pair of two-stage Kalman filters to detect and isolate failures in aircraft with redundant actuators

Abstract

As critically-safe systems, aircraft must be able to adapt if a failure occurs. A fast failure detection may reduce the possibility of loss of control in-flight. Model-based Failure Detection and Isolation Systems (FDIS) have been proposed to identify actuation failures in aircraft, being most of them based on Multiple Models approaches. An additional challenge in FDIS strategies for aircraft is the high level of actuation redundancy. This article proposes to use only a pair of Two-Stage Kalman Filters (TSKF) for detecting and isolating failures. A Feed-Forward Differential (FFD) is introduced to facilitate the fault isolation process, providing further information with low perturbation during flight. The TSKF results are interpreted by a Supervisory Loop, which automatically diagnoses the aircraft for actuation failures. The proposed solution is validated using the Boeing 747 GARTEUR RECOVER Benchmark Simulator. The proposed solution has an overall better response when compared to a multiple-models solution of the literature with an 85% reduction in the computational time.