Structured non-self approach for aircraft failure identification within a fault tolerance architecture

Abstract

ABSTRACTWithin an immunity-based architecture for aircraft fault detection, identification and evaluation, a structured, non-self approach has been designed and implemented to classify and quantify the type and severity of different aircraft actuators, sensors, structural components and engine failures. The methodology relies on a hierarchical multi-self strategy with heuristic selection of sub-selves and formulation of a mapping logic algorithm, in which specific detectors of specific selves are mapped against failures based on their capability to selectively capture the dynamic fingerprint of abnormal conditions in all their aspects. Immune negative and positive selection mechanisms have been used within the process. Data from a motion-based six-degrees-of-freedom flight simulator were used to evaluate the performance in terms of percentage identification rates for a set of 2D non-self projections under several upset conditions.