Aircraft Abnormal Conditions Detection, Identification, and Evaluation Using Innate and Adaptive Immune Systems Interaction

Abstract

Abnormal flight conditions play a major role in aircraft accidents frequently causing loss of control. To ensure aircraft operation safety in all situations, intelligent system monitoring and adaptation must rely on accurately detecting the presence of abnormal conditions as soon as they take place, identifying their root cause(s), estimating their nature and severity, and predicting their impact on the flight envelope.Due to the complexity and multidimensionality of the aircraft system under abnormal conditions, these requirements are extremely difficult to satisfy using existing analytical and/or statistical approaches. Moreover, current methodologies have addressed only isolated classes of abnormal conditions and a reduced number of aircraft dynamic parameters within a limited region of the flight envelope.This research effort aims at developing an integrated and comprehensive framework for the aircraft abnormal conditions detection, identification, and evaluation based on the artificial immune systems paradigm, which has the capability to address the complexity and multidimensionality issues related to aircraft systems.Within the proposed framework, a novel algorithm was developed for the abnormal conditions detection problem and extended to the abnormal conditions identification and evaluation. The algorithm and its extensions were inspired from the functionality of the biological dendritic cells (an important part of the innate immune system) and their interaction with the different components of the adaptive immune system. Immunity-based methodologies for re-assessing the flight envelope at post-failure and predicting the impact of the abnormal conditions on the performance and handling qualities are also proposed and investigated in this study.The generality of the approach makes it applicable to any system. Data for artificial immune system development were collected from flight tests of a supersonic research aircraft within a motion-based flight simulator. The abnormal conditions considered in this work include locked actuators (stabilator, aileron, rudder, and throttle), structural damage of the wing, horizontal tail, and vertical tail, malfunctioning sensors, and reduced engine effectiveness. The results of applying the proposed approach to this wide range of abnormal conditions show its high capability in detecting the abnormal conditions with zero false alarms and very high detection rates, correctly identifying the failed subsystem and evaluating the type and severity of the failure. The results also reveal that the post-failure flight envelope can be reasonably predicted within this framework. . . DEDICATION . DEDICATION To my parents, Zeki and Anwar, my wife, Rawaa, and my children, Ahmed and Yisra. iii . . ACKNOWLEDGMENTS . ACKNOWLEDGMENTSIt is a great pleasure to take this opportunity to thank my advisor, Dr. Mario G. Perhinschi. Without his true mentoring, indefinite support, and tireless patience, this work would not have been possible. I would also like to express my sincere thanks to the me...

Comparison of immunity-based schemes for aircraft failure detection and identification

“…The artificial DC mechanism proposed by the authors [45][46][47] is a computational algorithm inspired from the functionality of the biological DCs in the tissue and their interaction with T-cells and antibodies in the adaptive immune system. Cells from a pool of DCs are randomly selected to process the outcomes of the self/nonself discrimination process.…”

“…In particular, the peptides that appear on the surface of a mature DC are specific to the invading antigen [53]. Similarly, the migrated DCs in the artificial DC mechanism presented in Sec.IIIcarry information that is useful in identifying the failed subsystem [41][42][43][44][45][46][47]. This information is summarized by the triggered-features matrix, 1 F , defined in Eq.…”

“…These efforts have been focused primarily on aircraft systems fault detection and identification; however, they only have considered single classes and high magnitudes of failure for limited regions of the flight envelope. Therefore, the availability of failure detection and identification schemes with high rates of success, with comprehensive coverage, integrating all aircraft sub-systems and operational modes is a critical objective of current research efforts at West Virginia University (WVU) and Embry-Riddle Aeronautical University (ERAU) [36][37][38][39][40][41][42][43][44][45][46][47].…”

Application of an AIS to the problem of through life health management of remotely piloted aircraft

“…The limitation of this type of flight simulator is internal system faults cannot be learned due to the coarse nature of the model. [39] An aircraft manufacturer could use this technique however as they would likely possess a more detailed system architecture and dependency model created during the design and evaluation of their RPAS. The failure of the VPP servo actuator was modelled by approximating the force required to alter the pitch of the propeller at speed and then implementing a ramp input to gradually reduce the servo power until it could no longer move.…”