About the Project
AI_CDT_DecisionMaking
Details
Thermal management systems are essential to the performance, efficiency and reliability of fuel cells, batteries, power electronics and industrial energy systems. They must maintain critical component temperatures within safe limits while minimising cooling energy under changing heat loads and operating conditions. Important states, including internal temperatures, local heat generation and thermal resistances, often cannot be measured directly. Conventional controllers therefore rely on incomplete measurements and simplified models, limiting their response to uncertain and evolving behaviour.
Physics-informed machine learning can combine sensor data with thermodynamic and heat-transfer principles to estimate unmeasured states, while reinforcement learning can learn adaptive strategies that improve efficiency and transient response. Integrating them raises challenges in partial observability, model uncertainty, safety and generalisation. Existing approaches often treat estimation and control separately, without considering how estimated-state uncertainty should affect control or how actions could improve observability.
RQ1 State estimation: how can sparse measurements, conservation laws and dynamic thermal models reconstruct unmeasured temperatures, heat loads and parameters with calibrated uncertainty? RQ2 - Safe adaptive control: how can uncertainty-aware, safety-constrained RL optimise thermal regulation and parasitic energy while respecting temperature, pressure, flow and actuator constraints? RQ3 - Joint estimation and control: does integration improve safety, efficiency and robustness compared with independently developed functions, and can actions actively improve knowledge of the thermal state? RQ4 - Robustness and transfer: how can reliable performance be maintained under sensor loss, imperfect models and operating conditions beyond training?
The methodology will create representative dynamic models, potentially for liquid-cooled fuel-cell, battery or power-electronics systems, and generate synthetic and, where available, experimental data across heat loads, ambient conditions, faults and parameter variation. Physics-informed probabilistic estimators will combine observations with energy balances, heat-transfer equations and dynamic models. Candidate methods include physics-informed neural networks, Gaussian-process or Bayesian neural-network surrogates, differentiable state-space models and hybrid Kalman-filter/learning approaches. Outputs will include state predictions and calibrated uncertainty, with ablations testing the contribution of data and physics.
Safe adaptive controllers will use constrained or risk-sensitive RL. Safety layers, control-barrier or Lyapunov constraints, action shielding and fallback control will be investigated to prevent unsafe exploration. Estimated-state uncertainty will enter the policy or safety mechanism, making control more conservative when knowledge is weak. Dual-control behaviour will also be studied, allowing actions to improve regulation and information about uncertain states or parameters. Online updating, domain randomisation, transfer learning or robust policy design will address model drift, ageing, sensor loss and unseen regimes.
Performance will be benchmarked against Kalman-filter-based estimation, PID control and model predictive control, and against estimation and RL developed separately. Measures will include state-estimation error, uncertainty calibration, thermal-constraint violations, cooling or pumping energy, transient response, robustness to model mismatch and sensor loss, and generalisation. Targeted hardware-in-the-loop or experimental validation will demonstrate feasibility where appropriate. The outcome will be a transferable framework integrating probabilistic state estimation with safe RL for partially observable thermal systems.
Desirable Student Background:
A strong first degree or MSc in machine learning, computer science, control, mechanical or electrical engineering, applied mathematics, physics or a related quantitative subject. Python, mathematical modelling and numerical methods are important. RL, probabilistic ML, state estimation, heat transfer or model predictive control would be advantageous.
How to apply:
Please apply under the University of Manchester application portal and select 'PhD in Artificial Intelligence' (OAA Applicant Portal).
When applying, you’ll need to specify the full name of this project, the name of your supervisor, if you already having funding or if you wish to be considered for available funding through the university, details of your previous study, and names and contact details of two referees.
Your application will not be processed without all of the required documents submitted at the time of application, and we cannot accept responsibility for late or missed deadlines. Incomplete applications will not be considered.
You will be asked to upload the following supporting documents:
- Final Transcript and certificates of all awarded university level qualifications
- Interim Transcript of any university level qualifications in progress
- CV
- Supporting statement: A one or two page statement outlining your motivation to pursue postgraduate research and why you want to undertake postgraduate research at Manchester, any relevant research or work experience, the key findings of your previous research experience, and techniques and skills you’ve developed. (This is mandatory for all applicants and the application will be put on hold without it).
- Contact details for two referees (Please ensure referee email addresses are official university or workplace accounts, as we may need to verify the references.)
- English Language Certificate (if applicable)
We actively encourage applicants from diverse career paths and backgrounds and from all sections of the community, regardless of age, disability, ethnicity, gender, gender expression, sexual orientation and transgender status.
We also support applications from those returning from a career break or other roles. We are dedicated to supporting work-life balance and offer flexible working arrangements to accommodate individual needs. Our selection process is free from bias, and we are committed to ensuring fair and equal opportunities for all applicants.
We are dedicated to supporting work-life balance and offer flexible working arrangements to accommodate individual needs, including a part-time study option.
Join us and help grow the UK’s national capability in artificial intelligence, part of the UK’s modern industrial strategy. Everyone is welcome if you are passionate about uncertainty quantification, decision making with humans in the loop and decision making for machine learning systems and their applications in Science and Engineering, and looking to develop your research and innovation skills at the doctoral level to further your career in the AI sector.
Equality, diversity and inclusion is fundamental to the success of The University of Manchester, and is at the heart of all of our activities. We know that diversity strengthens our research community, leading to enhanced research creativity, productivity and quality, and societal and economic impact (Equality, diversity and inclusion | The University of Manchester).
We actively encourage applicants from diverse career paths and backgrounds and from all sections of the community, regardless of age, disability, ethnicity, gender, gender expression, sexual orientation and transgender status.
Funding Notes
Fully funded AI UKRI CDT 4 year program; The Project will be based in University of Manchester.
The UKRI AI CDT in Decision Making for Complex Systems is part of the TechFirst offering stipends of £30,000 for students eligible for home fee status. The start date is September 2026.
This project is with Qdot.
References
Raissi et al. (2019), Journal of Computational Physics, 378, 686-707. DOI: 10.1016/j.jcp.[contact details available in the full listing]
Garcia and Fernandez (2015), JMLR, 16, [contact details available in the full listing].
Mesbah (2018), Annual Reviews in Control, 45, 107-117.
Garg, Silvas and Willems (2025), Control Engineering Practice, 164, 106458.
Xing et al. (2026), Energy, article 142371.
