20505 research outputs found
Sort by
Physics informed generative design of invariant manifolds in the circular restricted three body problem
No full textThe increasing prevalence of small spacecraft platforms in Cis-Lunar space has underscored the need for computationally efficient and fuel-optimal trajectory design methods. Invariant manifolds in multi-body gravitational systems provide natural, low-energy pathways for space missions, but their computation remains reliant on numerically intensive techniques that require integration of the variational equations and sampling across large families of periodic orbits. These methods are poorly suited for onboard computation or rapid preliminary analysis. This thesis proposes a physics-informed generative framework for invariant manifold design in the Circular Restricted Three Body Problem (CR3BP), using a Vector Quantised Variational Autoencoder (VQ-VAE) trained on trajectory data associated with Earth–Moon L1 halo orbits. The model learns a discrete latent representation of the manifold geometry conditioned on physical parameters such as Jacobi constant and manifold branch. A physics-consistent loss is introduced to enforce compliance with the CR3BP dynamics during training. The results presented in this work demonstrate the feasibility of using generative deep learning models to approximate invariant manifold structures in the CR3BP. By learning a discrete latent representation of manifold trajectories directly from trajectory data, the proposed model provides a novel, data-driven perspective on a classic problem in astrodynamics. This work represents an initial step toward the use of generative models in astrodynamics. It offers a foundation for further exploration of latent-space representations, and highlights the potential for deep learning to support manifold discovery, low-energy transfer design, and the generation of physically consistent trajectories without reliance on traditional generating orbit methods.MSc in Astronautics and Space Engineerin
Federated Learning - Contextual Efficiency Optimisation and Adversary
No full textArtificial Intelligence (AI) has been widely studied, applied, and achieved impressive
successes in various sectors. Among these successes, Federated Learning (FL) stands
out as an innovative paradigm that enables distributed devices to collaboratively train
machine learning models without transferring sensitive data to central servers. This de centralised paradigm, crucial for privacy preservation, also introduces significant prac tical challenges, particularly due to the heterogeneity among participating clients, which
include variations in computational power, distinct data distributions, and diverse user
behaviours, all of which complicate the training process in efficiency and convergence.
Existing FL frameworks often apply a uniform learning configuration across all clients,
neglecting individual device characteristics and user contexts, inevitably leading to in efficient utilisation of hardware resources and increased energy consumption within the
federation. Moreover, optimisation techniques intended to exhaust computation potential
of the system, can inadvertently expose systems to vulnerabilities, compromising security
and reliability at both software and hardware levels.
This thesis addresses the critical gap of the contexts of FL including heterogeneous
hardware specifications and optimisation technique employments, by proposing adaptive,
context-aware methodologies that optimise efficiency, while analysing and mitigating as sociated security risks within federated learning environments. First, a Mixed-Precision
Over-the-Air Federated Learning (MP-OTA-FL) framework is introduced, allowing cli ents to operate at different precision levels based on their hardware capabilities. Exper imental results show that MP-OTA-FL significantly improves energy efficiency and task
performance, particularly for resource-constrained devices. Second, a novel Retrieval Augmented Generation (RAG)-based user profiling framework is proposed, dynamically
optimising client precision decisions based on contextual factors such as hardware con straints and user preferences, achieving higher global model accuracy and user satisfac tion. Finally, the thesis explores hardware vulnerabilities introduced by FL efficiency
optimisations, presenting a remote Rowhammer attack vector exploiting physically in duced perturbations on client devices, demonstrating practical security risks under real istic conditions. Collectively, these contributions provide comprehensive strategies for
contextually optimised, secure, and efficient FL deployment, balancing performance and security considerations. The findings offer foundational guidance for developing robust
FL systems tailored to diverse and resource-limited real-world scenarios.PhD in Aerospac
Design of a robust adaptive neural network MPPT controller for a photovoltaic energy storage system
No full textThis paper presents the design of an L1 Adaptive Neural Network (ANN) Maximum Power Point Tracking (MPPT) controller for a Photovoltaic Energy Storage (PV-ES) system consisting of a PV module, a DC–DC boost converter, and a battery storage unit. The controller combines the L1 adaptive control method with a Radial Basis Function (RBF) neural network to approximate unknown nonlinear dynamics and compensate for system uncertainties. This integration enables fast adaptation while preserving robustness, thereby overcoming key limitations of conventional MPPT strategies. Simulation results demonstrate that the proposed L1 ANN-MPPT controller ensures rapid convergence to the maximum power point, reduced steady-state oscillations, and enhanced battery charging efficiency under highly variable irradiance and temperature conditions. These findings quantitatively confirm that the proposed controller eliminates the conventional trade-off between dynamic response and steady-state precision, offering superior speed, stability, and accuracy in maximum power point tracking. Overall, the results validate its effectiveness and highlight its potential for real-time deployment in renewable energy systems.Journal of Control, Automation and Electrical System
Integration of concentrated solar power with solid oxide electrolysis for green hydrogen production: a comprehensive review
No full textThe integration of Concentrated Solar Power (CSP) and Solid Oxide Electrolysis (SOE) holds great promise for efficient and sustainable green hydrogen production. However, there is a lack of comprehensive studies reviewing the combined potential of these two technologies, which could offer enhanced efficiencies and reduced costs for large-scale hydrogen production. This review addresses that gap by analyzing the technical and economic feasibility of integrating CSP with SOE systems. This review provides a comprehensive analysis of the integration between CSP and SOE systems for green hydrogen production. The study examines critical technical challenges, including high operating temperatures, material compatibility, and heat transfer efficiency, while evaluating the economic feasibility of these integrated systems. Different CSP configurations are analysed based on their ability to provide heat alone or both heat and electricity, with thermal energy storage identified as a key factor in enhancing system performance by mitigating intermittency issues. Methodologies used in integration studies, such as simulation models and experimental setups, are critically reviewed, highlighting gaps in practical designs and real-world applications of CSP-SOE systems. However, despite these promising advances, only one laboratory-scale prototype has been demonstrated to date, underscoring the urgent need for pilot-scale CSP–SOE field testing under real direct normal irradiation (DNI) and thermal energy storage (TES) conditions. By addressing these technical and economic obstacles, this review offers insights into optimising CSP-SOE systems for sustainable, large-scale hydrogen production and provides actionable recommendations for future development.The authors extend their appreciation to the Deanship of Scientific Research at Northern Border University, Arar, KSA for funding this research work through the project number “NBU-SAFIR-2024”.Frontiers in Energy Researc
A three-stage evaluation of airline low-carbon competitiveness
No full textAchieving deep decarbonisation in aviation requires a systematic understanding of how airlines differ in their carbon efficiency and what operational or technological factors drive these differences. This study proposes an integrated three-stage analytical framework that combines a non-oriented SBM-DDF model, the Global Malmquist–Luenberger (GML) productivity index, Random Forest regression and a cloud-model heterogeneity assessment to evaluate the low-carbon competitiveness of 20 major global airlines from 2018 to 2022. The SBM-DDF model benchmarks multi-input–multi-output environmental efficiency, while the GML index captures dynamic productivity changes and decomposes them into efficiency change and technical change. Random Forest analysis identifies the key operational determinants of carbon efficiency, and the cloud model characterises heterogeneity in performance level, volatility and uncertainty.
Results reveal pronounced cross-airline and intertemporal heterogeneity. Mean efficiency declined markedly during 2020–2021 and rebounded in 2021–2022, although absolute CO2-slack reductions lagged behind relative technical efficiency improvements. GML analysis shows that productivity changes were modest and mainly driven by managerial and operational efficiency rather than technological progress, indicating the limited short-run impact of fleet renewal on low-carbon competitiveness. Airline-level patterns demonstrate consistently strong and stable performance among Singapore Airlines, EasyJet and Cathay Pacific. Random Forest results identify revenue tonne-kilometres (RTK), employment scale and fleet size as dominant drivers of emission efficiency, while the cloud-model heterogeneity typology reveals four distinct groups ranging from efficient-stable to inefficient-volatile. Policy implications emphasise the need for performance-sensitive benchmarks in ETS and SAF-crediting schemes, and for managerial efficiency improvements to complement long-run technological transitions.This study was supported by the National Key Research and Development Project (NO.2022YFB2602000), the National Nature Science Foundation of China (U2333217), and the Fundamental Research Funds of CASTC (XXX252060302025029).Transportation Research Part D: Transport and Environmen
Dataset "Analytical modelling for prediction and prevention of overflow occurrence in wire-based additive manufacturing"
No full textWire-based directed energy deposition (w-DED) is an emerging additive manufacturing technology that has great potential for more efficient building of
large-scale metallic structural components. Increasing the deposition rate is one of the primary goals to meet the low-cost demand for large-scale parts.
However, a high deposition rate usually requires higher heat input to fully melt the material, which could lead to overflow defects. The occurrence of overflow
can be attributed to those factors which interfere with the balance between gravitational and surface tension forces acting on the molten material. This
disruption reduces the melt pool stability and imbalance the forces, ultimately generating overflow defects in the deposited layer. This paper presents a
thermo-capillary-gravity model for predicting the overflow occurrence based on a criterion of the analytically calculated reciprocal Bond number, V80.
Experiments show that if the 1/B0 is no greater than 0.74 or the bead height is no less than 1.16 times the capillary length, there will be a high probability of
overflow occurrence. Two w-DED processes using mild steel and duplex stainless-steel wires were employed to validate the model, demonstrating an overall
average accuracy of 84% and 93%, respectively. It is also found that both energy and material feed rates, which are two generic physical factors, significantly
affect the molten material overflow. The validated modelling approach enables efficient prediction and control of overflow for high-quality high deposition
rate w-DED processes.Engineering and Physical Sciences Research Council (EPSRC)NEWAM programme (EP/R027218/1
Understanding the Airport City - An Analysis and Critique of Airport City Development
No full textCities have always formed around transport hubs, and airports are not exempt from this maxim. The last 20-30 years have seen an exponential increase in the development of urban areas around airports, known as airport cities. These ‘airport cities’ offer an exciting vision of future living. However, there are problems associated with these developments and in the rush to assess the phenomenon, a fundamental questioning of if and how they should exist has been overlooked. Therefore, the aim of this thesis was to critically assess how, if at all, airport cities should evolve through understanding and analysing existing airport-centred urban development. An examination of the minutiae of daily life in the airport city alongside strategic analysis offers a fresh approach to airport city research and enables conclusions to be made about the existence and development of airport-centred urban areas. To achieve this, two airport cities have been studied; the Circle at Zurich Airport and Munich Airport City, specifically the LabCampus development. Through in-depth analysis and comparison of the two cases, this thesis found that airport cities can be best understood as extensions of the airport and are primarily ‘cities’ for work – the airport’s unique advantages lending themselves well to this purpose. However, to transform into true ‘cities’, become more sustainable and overcome the challenges of airport-urban integration, a reconsideration of ‘airport cities’ is required. It was concluded that airport cities need to be located in the main passenger flows and be integrated with the transport interchange of the airport to develop a better urban atmosphere, the Munich Airport Centre being highlighted as a good example of this. This, and similar examples, should be further studied as small-scale but high-quality airport-centred urban development that focus on retail, commercial and transport activities seem to offer the most viable and attractive way for airport cities to evolve in the future.MSc in Airport Planning and Managemen
Effect of pre-harvest nitrogen on hormone profiles and dormancy break in potato tubers
No full textThis study tested whether pre-harvest nitrogen (N) supply alters the timing of dormancy release and the profiles of abscisic acid (ABA) and cytokinins (CKs) in apical buds of potato tubers. A field experiment was conducted with two cultivars, Brooke and VR808, receiving standard (100% N) and reduced (30% N) doses. After harvest, tubers were stored and assessed weekly for dormancy followed by collection of apical buds for hormone profiling and subsequent analysis using Liquid Chromatography–Tandem Mass Spectrometry (LC-MS/MS). Multiple PGRs were targeted; of these, only ABA and DZR were quantifiable and profiled across stages and treatments. Reduced nitrogen delayed dormancy break in both cultivars by around one week. Across the various dormancy stages, ABA in the apical buds remained stable and unaffected by nitrogen, indicating that a decline in ABA cannot always be associated with dormancy break. DZR showed cultivarand nitrogen-dependent patterns, with shifts evident in Brooke but not in VR808, suggesting that cytokinin (CK) mobilisation could be genotype-conditioned. Overall, the findings suggest that N could modulate dormancy via altered ABA sensitivity rather than ABA decline, while DZR-linked cytokinin dynamics appear genotype-conditioned; this study also confirms the feasibility of nitrogen optimisation as an effective, environmentally friendly tool to extend dormancy, aid storage scheduling, and reduce fertiliser use.MSc in Future Food Sustainabilit
Interpreting and enhancing decisions in autonomous navigation: a belief-desire-intention reinforcement learning (BDI-RL) approach
No full textExplaining autonomy is becoming a crucial factor in the design of trustworthy autonomous platforms in both transport and smart living sectors. Interpretable reinforcement learning (RL) is an emerging research area that aims to explain why an autonomous platform adopts an action or set of actions. However, the state-of-the-art has focused on the design of explainable tools as independent modules that are not involved in the decision-making process of the RL agent. In this paper, we propose a novel belief-desire-intention RL (BDI-RL) approach that incorporates the explainable module as a belief model that enhances the learning capabilities of the RL as well as actions interpretability. To this end, we combine the merits of Dyna-Q algorithm as backbone RL model and belief maps as explainable element. The combined contribution of these models provides a robust model that emulates better the reasoning process of humans by leveraging beliefs and online agent-environment interactions. Simulations experiments are conducted in a grid environment of different sizes and obstacles. Comparisons are also provided to show the benefits of the proposed methodology.This work was supported by the Royal Academy of Engineering and the Office of the Chief Science Adviser for National Security under the UK Intelligence Community Postdoctoral Research Fellowship programme.2025 IEEE International Conference on Systems, Man, and Cybernetics (SMC
Developing a Reduced 2D Model for Rotating Detonation Engine Simulations
No full textThis thesis develops a compact, reproducible two-dimensional finite-volume Euler solver in vectorized Python to support exploratory detonation studies. The flow solver uses MUSCL reconstruction with an HLLE approximate Riemann flux and a strong-stability-preserving Runge–Kutta time integrator, emphasizing portability and clear, testable software structure. Verification exercises include a quantitative Sod shock tube study against an exact Riemann solution and a qualitative circular blast symmetry check; additional timing sweeps compare a MATLAB baseline to the vectorized Python port. Thermochemistry is designed but not exercised in the CFD results: instead, a standalone JAX-based multi reactor module is validated against Cantera and benchmarked for scaling, demonstrating increasing speedups with problem size while maintaining parity in instantaneous rates and thermodynamic conversions. Together, the results indicate close agreement with the MATLAB reference on canonical tests and credible throughput gains for batched chemistry, with an explicit path to operator-split coupling. Limitations include the absence of in-line chemistry in the delivered solver and a focus on ideal-gas mixtures; future work targets robust operator-split integration and extended verification and validation.MSc in Astronautics and Space Engineerin