Kamal Abdulraheem

Schmidt AI in Science Fellow, Michigan Institute for Data Science

My research focus is on the fusion of Artificial Intelligence and Control theory to design and develop autonomous control systems capable of decision-making, prognosis, diagnosis, adaptation, and control. My research goal in this fellowship is to enhance AI/ML capabilities for control theory with applications to complex engineering systems like nuclear power plants (NPPs). It concerns applying AI/ML practices, including Reinforcement Learning (RL) and Deep Neural Networks to numerically solve complex models in control theory, which are challenging using traditional methods. In this fellowship, we are going to develop a novel hybrid controller of MPC-RL with uncertainty quantification (UQ) for microreactor applications that combine the best of both worlds: RL develops a model that learns from previous states, while MPC ensures RL is fully respecting the physical model constraints by resolving neural network extrapolation issues.

  • AI Mentor: Majdi Radaideh, Alex Gorodetsky, Aerospace Engineering, College of Engineering
  • Science Mentor: Brendan Kochunas, Nuclear Engineering and Radiological Science, College of Engineering
  • Research Theme: AI management of nuclear reactors