This thesis addresses the topic of fault-tolerant flight control system (FTFCS) design and focuses on its application to the Bell-205 helicopter. In this context, a fault detection, isolation, and accommodation (FDIA) system has been constructed using artificial neural networks (ANNs) and real flight test data (FTD). The construction of the ANN-based FDIA system considers all the feedback sensors but does not use any of the sensor measurements in the input space of the ANNs. This latter feature increases the reliability of detection and isolation of faults. Desktop simulations of the FDIA system have shown highly acceptable performance. Robust controllers have been designed for the lateral and longitudinal dynamics using the Hinfinity mixed-sensitivity approach. The controllers were then integrated with the aforementioned FDIA systems and tested in simulation. The inspection of various uncertainties, including those due to the presence of the FDIA in the feedback loop, indicated that mu-synthesis may give better results than the Hinfinity design. Therefore, an improved FTFCS system was designed using mu-synthesis. The functioning of the integrated systems is acceptable but their accuracy needs to be further verified on the nonlinear model. Use of mu-synthesis helped to identify areas for further improvement. In addition to the design work that was carried out in the thesis, a theoretical investigation was conducted to study the impact of the faults on the Algebraic Riccati Equations (AREs) that are normally solved when finding stabilizing Hinfinity controllers. Accordingly, a new FTFCS scheme which is based on solving a new set of AREs is proposed. The solutions of the Riccati equations are corrected adaptively. The controller has a particular structure and only certain parts of it need updating. The advantages of this approach include possible smooth transfer when updating and less computations when compared with switched controllers.
展开▼
