https://www.um.edu.mt/library/oar/handle/123456789/66053 2026-04-04T17:43:01Z https://www.um.edu.mt/library/oar/handle/123456789/135060 Title: Anomaly detection and analysis of flight data using machine learning Abstract: This work and its abstract are both under embargo until the restriction is lifted. Description: Ph.D.(Melit.) 2024-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/132779 Title: On engineless taxiing with autonomous electric tow trucks Abstract: The increase in air traffic during the last decades has had a significant negative impact on the environment in terms of noise pollution, air pollution, and fuel emissions. Historically, efforts to increase efficiency have mostly concentrated on the airborne phase of the flight mission; however, recently, the ground phase of the flight mission, or taxiing, has been getting more attention. An aircraft's engines are optimised for cruise speed, hence using them for taxiing is rather inefficient. In addition, high traffic volumes and inefficient taxi operations – with frequent stop-and-go aircraft movements – lead to increased fuel consumption during taxiing. Thus, it is understood that one of the main sources of noise and pollution at airports is taxiing. This is one of the issues that the Single European Sky ATM Research Joint Undertaking (SESAR JU) initiative is tackling, and a number of options have been proposed as alternatives to conventional taxiing. One of the solutions proposed by the aerospace industry is to introduce electric tow trucks to tow aircraft from the stand to the runway (or vice-versa). However, the introduction of tow trucks results in more surface traffic, which is undesirable from the perspective of an Air Traffic Controller (ATCO), as it leads to higher workload. There are numerous ways to compensate for this, including the use of automated planning and execution, but the majority of the solutions proposed in the literature have one or more of the following drawbacks: severe limitations in the deviations that can be introduced in terms of planned taxi routes and scheduled timings; inability to schedule and plan routes for multiple active runways; and no consideration for tow-truck battery state-of-charge during planning. In terms of performance testing of such solutions, only singular performance metrics (e.g., number of vehicle conflicts or average taxi time) have been considered in the literature, thus limiting the validity and applicability of these solutions. To enhance ground operations and get around some of the drawbacks of current methods, this work details a novel algorithm for taxi operations using autonomous tow trucks. The algorithm identifies conflict-free solutions that limit taxi-related delays and fuel consumption while taking full advantage of the use of tow trucks for taxi operations. It can cater for multiple active runways and accounts for tow truck battery state-of-charge, as well as limits in the number of tow trucks, tow truck depots and charging stations. The algorithm operates at a strategic level (i.e. prior to the start of taxi operations) and uses a centralised approach (i.e. the algorithm is executed on a single computer and may be used as an Air Traffic Control (ATC) tool for ensuring adequate traffic separation at all times). The algorithm can use one of two approaches: a Time-Wise Approach – which prioritises taxi delays over fuel consumption – and a Fuel-Wise Approach – which prioritises fuel consumption over taxi delays. Additionally, when assigning tow trucks to arriving or departing aircraft, the algorithm can either use Static Allocation, in which case a tow truck must be parked in a depot in order to be assigned to an aircraft, or Dynamic Allocation, in which case tow trucks can be anywhere on the airfield when they are assigned to an aircraft. The proposed algorithm was tested for different airports with different numbers of active runways, with various levels of traffic and different quantities of tow trucks. A large number of performance metrics were defined to evaluate the performance of the algorithm. A realtime simulator was developed in order to test the updated schedules as provided by the algorithm, both in a deterministic environment (referred to as the Deterministic Model) – to ensure correct algorithm operation in terms of conflict-free routes – and in a probabilistic environment (referred to as the Probabilistic Model) – to test the robustness of the solutions when the updated schedule is executed under real-life conditions including uncertainty which results in deviations from the assumed parameters. When evaluated with the Deterministic Model, the results demonstrate that the algorithm is capable of using tow trucks for aircraft taxiing without causing any traffic conflicts; to achieve conflict-free routes, nearly 70% of the flights required modest delays of up to 3 minutes to ensure that adequate traffic separation was maintained at all times. These delays were mostly caused by waiting at the stand before taxiing (for departures) or near the runway (for arrivals), and to a lesser extent caused by deviations from an ideal (i.e. shortest) taxi route. The algorithm is capable of assigning tow trucks to more than 90% of the flights, even with a tow truck fleet as small as 30% of the airfield’s hourly rate of traffic, with small differences observed for different algorithm settings. In the Probabilistic Model, uncertainties in the vehicle velocities and in the aircraft start time are introduced. When tested with this model, the algorithm shows higher robustness with Static Allocation than with Dynamic Allocation, in particular in terms of delays and number of tow trucks which fail to attach to the assigned aircraft, therefore indicating that the Dynamic Allocation settings do not always have a favourable impact on the algorithm, especially when there are large margins of uncertainty. The algorithm was also tested for different battery performance characteristics. For relatively small variations in the velocity of discharge rates, significant variations in tow truck performance and average fuel savings were observed. This emphasises the impact of battery characteristics on tow truck performance, and on the size of the tow truck fleet that would be required to cope with the ground movements at a particular airfield. Description: Ph.D.(Melit.) 2024-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/118587 Title: Design of a flywheel for a kinetic energy recovery and storage system from a landing aircraft Abstract: With the ever-increasing environmental pressure, the aviation industry is in constant pursuit to reduce fuel consumption, both in flight as well as on the ground. This thesis addresses engineless aircraft ground movement by developing a concept for energy recovery and storage from a landing aircraft. Following an economic assessment to assess the concept, the thesis develops a design process to establish the shape, material, geometry and operational requirements of the flywheel. Following analytical calculations of the stresses induced in the flywheel, a finite element analysis of the flywheel under operational loads is executed and results analysed with reference to the suitability of the design method followed. Results illustrate that stresses induced by operational rotational speeds are within material stress constraints determined by the material properties. Description: M.Sc.(Melit.) 2021-01-01T00:00:00Z https://www.um.edu.mt/library/oar/handle/123456789/118584 Title: Aerodynamic stall recovery using artificial intelligence techniques Abstract: According to International Civil Aviation Organization (ICAO) in 2019 just before Covid-19 started, 4.5 billion people worldwide travelled by plane. That is nearly half the population of the world. Although, accidents happen, statistically, it is the safest mode to travel from one place to another as of now. This is because of extensive training given to pilots, the on-board automation and protection systems of modern commercial aircraft. Despite the on-board automation and protection systems, aerodynamic stall is still a possible occurrence and pilots undergo stall detection and recovery training to deal with such scenarios. Nevertheless, accidents have occurred due to pilot error during stall recovery. This work uses combination of Reinforcement Learning (RL) algorithm, Behavioral Cloning (BC) and Deep Learning (DeL) based regression model to train multiple Machine Learning (ML) models to automatically recover an aircraft from a wings level (1G) stall, stall during a turn and stabilize it. The RL environment consists of X-Plane flight simulator, NASA’s XPlaneConnect plugin to interface X-Plane with python programming language. The design of whole setup and implementation of the algorithms is discussed, together with the training and testing of the ML models in this dissertation. The stall recovery process was divided into two parts. The first part was about reducing the Angle of Attack (AoA) below the critical AoA. Once, the current AoA is less than critical AoA then the side stick controls are handed over to the second ML model. The first agent is based on Deep Deterministic Policy Gradient (DDPG) algorithm, which has been pre-trained using BC technique. The data which is used to pre-train the actor network of DDPG is recorded with the help of expert pilots. Pre-training the actor network helps the RL agent to converge at a faster rate and learn a policy which is similar to that of an expert pilot. The pre-training of the actor network significantly reduced the time taken to train the first agent to recover from a stall as the actor network performed like an expert pilot. Once, the agent learns a policy which is similar to expert’s policy. A random noise is added to the output of the network to help the agent to explore the RL environment. This exploration is needed to help the agent to find better policies to recover from an aerodynamic stall. The second agent is a DeL regression model, which is trained on a different expert recorded dataset. The DL regression model is responsible to stabilize the aircraft and reach a safe airspeed once the RL model hands over the side stick control to the regression model. The results obtained in this work are satisfactory as the ML models have been able to recover from stall events at various altitudes between 3,000 feet and 30,000 feet. This dissertation present and discuss the training and test results in further detail and will examine the sensitivity of the algorithms to various other factors. Description: M.Sc. (Melit.) 2021-01-01T00:00:00Z