Controls Research Group

Research Focus Areas

Digital Twins

We establish general mathematical and computational frameworks for real-time digital twins, with applications demonstrated on nuclear power plant systems.

Intelligent Control

We leverage deep reinforcement learning to develop real-time, uncertainty-aware control strategies for both light water and advanced microreactor concepts.

Classical Control Theory

We apply both linear and nonlinear model predictive control techniques, and integrate them with reinforcement learning to create hybrid algorithms.

Anomaly Detection

We develop data-driven methods to identify anomalies and predict potential faults in sensor data.

Robotics

We program drones and quadruped robots for autonomous inspection tasks and radiation mapping in nuclear power plant environments.

Recent Publications

A load following reactivity control system for nuclear microreactors

2025

Progress in Nuclear Energy

Multistep Criticality Search and Power Shaping in Nuclear Microreactors with Deep Reinforcement Learning

2025

Nuclear Science and Engineering

Distance preserving machine learning for uncertainty aware accelerator capacitance predictions

2024

Machine Learning: Science and Technology

View Paper

Simplified matching pursuits applied to 3D nuclear reactor temperature distribution construction

2024

Applied Mathematical Modelling

View Paper

Operational data for fault prognosis in particle accelerators with machine learning

2023

Data in Brief

View Paper

Recent Funded and Ongoing Projects

•Generative diffusion models for full-field reconstruction in nuclear digital twins
(Further read:
Work in progress
)
•Collaborative drone-based autonomous inspection in nuclear-plant-like environments
(Further read:
Work in progress
)
•Real-time sensor data assimilation into digital twins using variati onal inference techniques
(Further read:
Work in progress
)
•Deep reinforcement learning for real-time control of nuclear reactors
(Further read: https://arxiv.org/abs/2504.00156)
•Classical control a pproaches for load-following system design in nuclear microreactors
(Further read: https://doi.org/10.1016/j.pnucene.2025.105676)
•Recurrent neural networks and autoencoders for anomaly detection in electronic power signals
(Further read: https://doi.org/10.1016/j.dsp.2022.103704)
•Fault prognosis in particle accelerators using ensemble learning techniques
(Further read: https://doi.org/10.36001/ijphm.2023.v14i1.3419)
•Full-field temperature reconstruction in microreactor systems via matching pursuits
(Further read: https://doi.org/10.1016/j.apm.2024.04.011)
•Reactivity control and criticality search using swarm intelligence and evolutionary algorithms
(Further read: https://doi.org/10.1016/j.nucengdes.2022.111776)
•Forward digital twins for monitoring nuclear reactor transients and accident progression
(Further read: https://doi.org/10.1016/j.eswa.2020.113699)