Fault Detection, Identification and Accommodation Techniques for Unmanned Airborne Vehicles

摘要

Unmanned Airborne Vehicles (UAV) are assuming prominent roles in both the commercial and military aerospace industries. The promise of reduced costs and reduced risk to human life is one of their major attractions, however these low-cost systems are yet to gain acceptance as a safe alternate to manned solutions. The absence of a thinking, observing, reacting and decision making pilot reduces the UAVs capability of managing adverse situations such as faults and failures.ududThis paper presents a review of techniques that can be used to track the system health onboard a UAV. The review is based on a year long literature review aimed at identifying approaches suitable for combating the low reliability and high attrition rates of today’s UAV. This research primarily focuses on real-time, onboard implementations for generating accurate estimations of aircraft health for fault accommodation and mission management (change of mission objectives due to deterioration in aircraft health).ududThe major task of such systems is the process of detection, identification and accommodation of faults and failures (FDIA). A number of approaches exist, of which model-based techniques show particular promise. Model-based approaches use analytical redundancy to generate residuals for the aircraft parameters that can be used to indicate the occurrence of a fault or failure. Actions such as switching between redundant components or modifying control laws can then be taken to accommodate the fault.ududThe paper further describes recent work in evaluating neural-network approaches to sensor failure detection and identification (SFDI). The results of simulations with a variety of sensor failures, based on a Matlab non-linear aircraft model are presented and discussed. Suggestions for improvements are made based on the limitations of this neural network approach with the aim of including a broader range of failures, while still maintaining an accurate model in the presence of these failures.