The availability of free satellite images in medium/high special resolution has enabled possible solutions for the challenging dynamic land use and land cover mapping problems. This project is about developing new AI techniques for crop detection and mapping. Crop detection is the first step in AI-based time series analyses, aiming to provide fundamental information for many socio-economic applications. Examples are crop control and yield estimation, change monitoring, supply chain and food security, climate change policies such as crop rotation, insurance, and fertilization services. The lack of ground truth data is a major problem for crop detection. That is the case for most time-series analyses of historical data. On the other hand, the crop-specific variations in visual and chemical characteristics during a year are sensed by spectral satellite images. Therefore, this project is focused on developing AI techniques effectively utilizing the spectral bands for crop detection. This is achieved based on (1) an unsupervised framework to identify effective spectral bands for time-series analysis and crop detection. For this aim, the crops fingerprints and other vegetation indexes are used to identify the important spectral wavelengths. (2) a supervised framework for developing a novel spectral attention model using visual transformers prediction strategies.

One of the major knowledge gaps in the study of complex aerospace systems is the frictional interface of jointed components. In the current state-of-the-art, only the nonlinear friction is considered. For the underlying physics, frictional energy dissipates energy via deformations (using current methods), heat (not considered and is the focus of this project), and sound (considered ignorable for most sliding situations). This project will focus on the understanding of the heat-based energy dissipation and the environmental effects on these frictional joints. Swansea University is a member of the International Committee on Joint Mechanics. This is a multi-disciplinary committee focused on understanding and predicting jointed interfaces in assembled structures. The ICJM is a collection of academics, industrial researchers, and governmental bodies from across the world. This allows for knowledge transfer between industrial needs, current state-of-the-art, and relivant regulations. The successful applicant will have the opportunity to attend joint community meetings, where they can discuss and present their work, as well as attending and presenting at international conferences. This project will investigate the recently identified need (discovered by the ICJM) for multi-physics understanding and modelling. Specifically, this project will investigate the thermal-mechanical relationship in assembed structures. The student will utilise a newly acquired environmental chamber to understand these previously not investigated aspects of these nonlinear joints. In addition to the testing, the student will also work towards developing a novel temperature-dependent nonlinear joint model.

The primary goal of this research project is to enhance the safety and comfortability of High-Speed Railway (HSR) infrastructures by addressing key challenges associated with their maintenance and operation during regular public transportation service. The project aims to answer the following critical questions: 1) How can deep learning techniques be applied to detect and assess structural distresses in High-speed Railway infrastructures, such as cracks in concrete slabs, more accurately and efficiently? 2) What are the best practices for using deep learning to monitor and predict foundation settlements that may compromise the stability and comfort of HSR systems? During the course of the project, an innovative integration of advanced deep learning techniques, including computer vision, Large Language Models (LLMs), etc. will be developed to address the challenges in HSR infrastructure engineering monitoring and maintenance: 1) Deep learning methods: This involves the development of algorithms capable of analysing images and videos of railway infrastructures to detect and quantify cracks, deformations, and other structural issues. 2) Large Language Models (LLMs): These models will be used to process and analyze vast amounts of vibration and settlement data related to HSR infrastructures under long-term service. Students involved in this project will undertake a variety of tasks, including: 1) Data Collection and Pre-processing: Gathering and preparing large datasets from various sources, including field inspections, historical maintenance records, and sensor data. 2) Algorithm Development: Designing and implementing deep learning models for detecting structural issues and predicting future maintenance needs. This will involve programming, model training, and fine-tuning. 3) Simulation and Analysis: Using deep learning methods to simulate the behavior of High-speed railway infrastructures under different conditions and validate the models against real-world data. 4) Integration and Testing: Developing an integrated system, and testing this system in a controlled environment before deployment on actual HSR infrastructure systems. 5) Reporting and Presentation: Documenting findings, preparing technical reports, and presenting results to stakeholders and the broader scientific community. This project not only aims to improve the safety and comfort of HSR systems but also seeks to provide students with hands-on experience in cutting-edge technologies, preparing them for future careers in engineering and data science.

For optimal design of engineering structures it is important to consider uncertainties in specifying system parameters, boundary conditions and applied loading. In safety-based optimal design the effect of uncertainties are explicitly considered at the design stage, which is not the case in conventional design methods. Failure to consider uncertainty can lead to unreliable, uneconomical and even unsafe products, proving costly to the industries, and indeed to the economy in general. This proposal outlines a five-year work-programme aimed at the development of safety-based optimal design tools for engineering structures subjected to a wide range of dynamic loading. The methods available to handle uncertainties in structural dynamics can be broadly divided into two groups: (a) the methods applicable for low-frequency vibration (e.g., Finite Element (FE) method) and (b) methods applicable for high-frequency vibration (e.g., Statistical Energy Analysis (SEA)). The developments of these two groups of methods have tended to take place independently with little overlap between them. Up until now there is no method suitable for mid-frequency vibration, which is important in many application areas, for example, in aerospace and automotive industries. The proposed research will bridge this gap by going from the 'low-frequency end' to the 'high-frequency end' and the new methods will be integrated with the optimal design process. The overall outcome of the project will be numerically and experimentally validated unified design tools that can be used to optimally design dynamic engineering structures meeting a priori prescribed safety targets. Proposed work would also integrate the newly developed tools with existing industry standard design tools (commercial FE software) so that it can be incorporated easily within the existing design facilities without significant additional investments. The benefits of probabilistic design methods are yet to be fully appreciated in many industrial sectors and the success of the proposed project will be a motivation to move away from the paradigm of traditional safety factor based design philosophy.

This PhD project aims to enhance the digital twinning process for fusion energy components by leveraging physics-based modelling, data analysis, and artificial intelligence (AI) techniques. The primary objective is to accelerate the digital twinning loop, which involves continuously refining and improving digital twin models based on real-world data and feedback. In addition to reviewing physics-informed neural networks and standard finite element modelling, the project will also investigate the possibility of improving finite element-based analysis by embedding some modern AI principles. This could involve manipulating the FE shape function using neural networks to conserve the formulation's energy. The PhD project ensures that all applications investigated are directly relevant to the challenges faced by the UK Atomic Energy Authority (UKAEA) in fusion energy research. This involves addressing issues related to component design, performance optimization, safety considerations, and other relevant aspects within fusion energy systems.