The Lincoln AI Laboratory (LAIL), part of the School of Engineering and Physical Sciences at the University of Lincoln, is specialised in developing cutting-edge AI-based solutions that enable industries to operate more reliably, efficiently, and sustainably. We work at the intersection of academia and industry.
Explore Our ResearchNews
Can DL accurately predict the flow field and aerodynamic performance of multi-stage axial compressors?
Next week, at ASME TurboExpo2025 Giuseppe will answer these questions, and provide our experience by sharing some the work we carried out at Siemens Energy, in collaboration with the Lincoln AI Lab (LAIL).
Read our paper here…
Our students attended a 2-week CDT workshop organised by Siemens Energy
Afila and Kenechukwu from LAIL joined students from the University of Cambridge to attend a 2-week CDT workshop by Siemens Energy. During the event, they gained hands-on experience in gas turbine technology, exploring topics like combustion, blade cooling, machine learning applications, and decarbonisation, while working on mini-projects guided by industry experts from Siemens Energy.
Kenechukwu wins Siemens Energy competition
Kenechukwu from LAIL, and Zain Robson from the University of Cambridgge won the Siemens Energy CDT comptetition. The competition investigated a current business challenge. Kenechukwu and Zain proposed the implementation of wall and dump exit suction in other to shorten the length of the secondary diffuser while maintain pressure recovery. Their proposed improvements to the secondary diffuser in gas turbine packages earned a narrow win.
Our Team
Our diverse team of researchers, engineers, and visionaries are united by a shared passion for pushing the boundaries of artificial intelligence
Dr. Sepehr Maleki
Lab Director
Sepehr is a Senior Lecturer in AI and Industrial Digitalisation at the School of Engineering and Physical Sciences. He is currently chairing the Digitalisation Research Network at the University. His research area includes AI, Industrial Digitalisation, Manufacturing, and algebraic methods for Fault Detection and Isolation in Industrial Systems.
Giuseppe Bruni
Giuseppe is a PhD student at LAIL, specialising in Machine Learning Modelling of Manufacturing Variations on Compressor Aerodynamic Performance. With extensive industrial background, Giuseppe played a pivotal role in establishing the vision and direction of the lab, becoming the second member to join this research group. Giuseppe is a Chartered Engineer with over ten years of expertise in aeromechanical analyses.
Beth Hallam
Beth is a PhD student at LAIL focused on designing custom attention models for time-series anomaly detection.
Hoda Elmi
Hoda is a PhD student at LAIL. Her research focuses on the application of reinforcement learning for enhancing anomaly detection in diffusion models through data masking strategies.
Shafagh Pashaki
Shafagh is a PhD student at LAIL. Her research aims to develop reinforcement learning-based approaches for the optimisation of demand response in power grid networks.
Afila A. S.
Afila is a PhD student at LAIL. Her research aims to develop Physics-Informed Neural Networks to enhance the design and optimisation processes for gas turbines – overcoming the cost barrier of traditional CFD methods.
Kenechukwu Ogbuagu
Kenechukwu is a PhD student at LAIL. His research is focused on Generative AI methods for facilitating testing of gas turbines. Previously, he was a research engineer at LAIL working on flow field prediction using diffusion models.
Taraneh Latifi Seresht
Taraneh obtained an MMath degree from the University of Edinburgh. Her research at LAIL focuses on accelerating diffusion models, leveraging mathematical optimisation and machine learning to enhance their efficiency and scalability.
Oladeji Olanrele
Oladeji is a PhD student at LAIL. His research is focused on machine learning-based approaches for production planning and optimisation of the fast-moving consumer goods. His research explores data-driven strategies to optimise supply chains and streamline manufacturing processes.
Research Projects
Our current research portfolio spans fundamental AI research to practical applications that will revolutionize how we interact with intelligent systems
Deep Learning for Physical Simulations
The drive toward sustainability and heightened performance in modern turbomachinery (particularly for aviation and energy generation) demands ever more compact, efficient, and lighter-weight engine architectures. Traditional conservative design strategies, while safe, limit the potential for fully exploiting new architectures and materials. This challenge is magnified by the vast multi-physics complexity inherent in turbomachinery, where aerodynamics and structural mechanics intersect, making full-scale experimentation and high-fidelity simulations computationally prohibitive at the earliest design stages.
Applications of deep learning to physical simulations, such as Computational Fluid Dynamics (CFD), have become an emerging research area. Multi-stage axial compressors for gas turbine applications represent a remarkably challenging case for such models due to the high-dimensionality of the regression of the flow-field from geometrical and operational variables. One of the key research areas at LAIL is the development of deep learning-based methodologies, as well as Physics-Informed Neural Networks (PINNs), to accelerate the design process while maintaining reliability.
Anomaly Detection
The need for robust and efficient anomaly detection in time series data is critical across industries such as manufacturing, energy, and aerospace, where early identification of irregular patterns can prevent costly failures and optimise system performance. Traditional statistical methods and rule-based approaches often struggle with the high-dimensional, noisy, and dynamic nature of time series data, particularly in complex systems like industrial machinery or power grids. Manual inspection and conventional machine learning techniques are frequently computationally intensive or lack the adaptability required for real-time applications. At LAIL, one of our key research areas is the development of advanced deep learning-based methodologies to enhance anomaly detection in time series data while ensuring scalability and reliability. These include reinforcement learning, which optimises detection strategies by learning adaptive policies in dynamic environments; transformer-based models, which leverage attention mechanisms to capture long-range dependencies and contextual patterns in sequential data; and diffusion models, which excel at modeling complex data distributions to identify subtle anomalies in multi-variate data. Our research aims to deliver scalable, real-time anomaly detection solutions that reduce false positives, enhance predictive maintenance, and enable proactive decision-making in complex industrial systems.
Demand Response Optimisation
The push for energy efficiency and grid stability in the residential sector has made demand response optimisation a critical area of focus, particularly as renewable energy integration and smart home technologies grow. Traditional demand response strategies, often reliant on static rule-based systems or simplified models, struggle to adapt to the dynamic and heterogeneous nature of residential energy consumption patterns, which are influenced by occupant behavior, appliance variability, and fluctuating grid demands. These challenges are compounded by the need to balance consumer comfort, cost savings, and grid reliability in real time, making conventional approaches computationally intensive or insufficiently flexible. At LAIL, one of our research areas is the development of reinforcement learning-based methodologies to optimise demand response in the residential sector while ensuring scalability and adaptability. These approaches leverage reinforcement learning to model complex, dynamic interactions between households and the grid, enabling adaptive policies that optimise energy usage, reduce peak loads, and respond to real-time pricing or renewable availability.
Generative AI methods for Gas turbine testing
The pursuit of enhanced performance and durability in gas turbines, critical for aviation and power generation, necessitates advanced testing methodologies to accurately assess thermal behavior under extreme operating conditions. Traditional testing approaches, such as physical instrumentation or computational simulations, often face limitations due to the complex, high-temperature environments within gas turbines, where heat transfer, material stress, and fluid dynamics interact dynamically. Temperature-sensitive paints (TSPs), which fluoresce in response to temperature changes, offer a promising non-intrusive method for mapping surface temperatures, but their application is challenged by the need for precise calibration, data interpretation, and handling of noisy or incomplete measurements in harsh environments. At LAIL, we develop Gen AI-based methodologies to enhance the testing of gas turbines, with a focus on optimising the use of temperature-sensitive paints. The aim is to reconstruct high-fidelity temperature distributions from sparse or noisy TSP data, enabling accurate thermal mapping without extensive physical instrumentation. Diffusion models can model complex thermal patterns, predict missing data points, and improve the robustness of TSP-based measurements by learning the underlying physics and sensor characteristics. By integrating these generative AI strategies, LAIL aims to deliver scalable, precise, and cost-effective testing solutions that enhance the reliability of thermal assessments, improve gas turbine design, and accelerate the development of next-generation, high-performance turbine systems.
Our Research Partner
Interested in pursuing a PhD with us? Get in touch!
Contact Us!