Neural Network Based UAV Fault Tolerant Control with Aerodynamic Disturbance Estimation and Compensation

In this thesis, the problem of controlling a quadrotor UAV is considered. It is done by presenting an original control system, designed as a combination of Neural Networks and Disturbance Observer, using a composite learning approach for a system of the second order, which is a novel methodology in literature. After a brief introduction about the quadrotors, the concepts needed to understand the controller are presented, such as the main notions of advanced control, the basic structure and design of a Neural Network, the modeling of a quadrotor and its dynamics. The full simulator, developed on the MATLAB Simulink environment, used throughout the whole thesis, is also shown. For the guidance and control purposes, a Sliding Mode Controller, used as a reference, it is firstly introduced, and its theory and implementation on the simulator are illustrated. Finally the original controller is introduced, through its novel formulation, and implementation on the model. The effectiveness and robustness of the two controllers are then proven by extensive simulations in all different conditions of external disturbance and faults.

Experimental analysis of the aerodynamic and mixing performance of a ventilation device

Numerous types of acute respiratory failure are routinely treated using non-invasive ventilatory support (NIV). Its efficacy is well documented: NIV lowers intubation and death rates in various respiratory disorders. It can be delivered by means of face masks or head helmets. Currently the scientific community’s interest about NIV helmets is mostly focused on optimising the mixing between CO2 and clean air and on improving patient comfort. To this end, fluid dynamic analysis plays a particularly important role and a two- pronged approach is frequently employed. While on one hand numerical simulations provide information about the entire flow field and different geometries, they exhibit require huge temporal and computational resources. Experiments on the other hand help to validate simulations and provide results with a much smaller time investment and thus remain at the core of research in fluid dynamics. The aim of this thesis work was to develop a flow bench and to utilise it for the analysis of NIV helmets. A flow test bench and an instrumented mannequin were successfully designed, produced and put into use. Experiments were performed to characterise the helmet interface in terms of pressure drop and flow rate drop over different inlet flow rates and outlet pressure set points. Velocity measurements by means of Particle Image Velocimetry were performed. Pressure drop and flow rate characteristics from experiments were contrasted with CFD data and sufficient agreement was observed between both numerical and experimental results. PIV studies permitted qualitative and quantitative comparisons with numerical simulation data and offered a clear picture of the internal flow behaviour, aiding the identification of coherent flow features.