Dynamic flight envelope assessment with flight safety applications.

Abstract

Aircraft have a manufacturer prescribed operating flight envelope for safe operation. exceeding these limits can result in unrecoverable departures or even structural failure. Numerous commercial aircraft accidents in the past have been attributed to loss-of-control (LOC) resulting from exceeding the safe operating flight envelope. Hence, real-time knowledge of the safe operating flight envelope is essential for safe flight operation, a problem known as dynamic flight envelope assessment. This dissertation explores dynamic flight envelope assessment from a control theoretic perspective. Two notions of the flight envelope, namely, the reachable sets and the region-of-attraction analysis are investigated. The NASA generic transport model (GTM) aircraft dynamics is used as an application problem. Linear and nonlinear techniques for flight envelope assessment are formulated in the linear matrix inequality (LMI) and sum-of-squares (SOS) framework, respectively. LMI and SOS problems are computationally tractable convex optimization problems for which many semi-definite programming solvers are available. This thesis also investigated fault detection and isolation strategies. Commercial jet transport aircrafts make extensive use of active controls. Faults or failures in the flight control system (FCS) elements like sensors or control effectors can lead to catastrophic failure. Model-based fault detection and isolation (FDI) filters can provide analytical redundancy by reliably detecting such faults in the system. Practical application of model-based FDI filters is limited so far due to poor performance, false alarms and missed detection arising out of uncertain dynamics of the aircraft, effect of nonlinearities in the system and the influence of closed-loop controllers. An application of closed-loop metrics to assess worst case FDI filter performance in the presence of a controller and uncertain dynamics is presented. Longitudinal GTM dynamics are considered. An H∞ FDI filter and a geometric filter design are compared using the metrics and the results validated through simulation. This research was expanded to include synthesis, real-time implementation and flight validation of robust FDI filters for a small uninhabited aerial vehicle. The influence of different closed-loop controllers on FDI filter performance is investigated. A comparison is presented between simulation of the predicted FDI filter performance and flight experiment results.