Fault tolerant control of aerospace systems using sliding modes

Abstract

The combination of sliding mode control and control allocation has proven to be a popular choice for fault tolerant control of aerospace systems, due to it's inherent robustness and its ability to reconfigure independently of the closed-loop system dynamics. This thesis expands on this methodology to consider the case of aerospace systems with actuator constraints, presenting two distinct paradigms for preventing saturation: control signal redistribution and interpolated control laws. It is demonstrated throughout that in certain cases, severe enough faults/failures can lead to situations where more typical sliding mode schemes saturate and as a result become unstable.

For the majority of this thesis the proposed control schemes are tested on a non-linear model of a blended wing body aircraft, chosen due to its highly redundant and coupled actuator suite and novel dynamics. The technical contributions of this work can be split into three categories:

-The development of a high-fidelity blended wing body aircraft model.

-Control allocation schemes which aim to prevent saturation through redistribution. \item Interpolated sliding mode control schemes which aim to prevent saturation through reducing closed-loop performance.

Throughout, rigorous stability proofs are conducted which demonstrate the various schemes robustness to a set of faults/failures and uncertainty introduced into the closed-loop through the imprecise reconstruction of the faults/failures. Several computationally lightweight optimisation schemes are presented which are demonstrated to be theoretically able to run on-line, towards the end of the thesis the implementability of one of the schemes is tested on a Parrot Rolling Spider drone during flight tests.