AI/ML have achieved great success in image processing, pattern recognition, and textual domains, but they still lack the accuracy, reliability and interpretability that are central for safety-critical and “precious data”-limited scientific discovery and engineering applications. It is important to embed expert knowledge, first principles, and physics in data-driven models as we aim to extrapolate for inverse design of new molecules and materials, and elucidate complex systems behavior. To that end, physics-constrained machine learning (PCML) has recently emerged as a powerful technique to ‘hard’ embed first principles-based domain knowledge directly into data-driven learning processes. In this talk, I will describe how we are advancing such methods for hybrid modeling, simulation, optimization and control, and explore what we have already achieved and what lies ahead. I will draw examples from a range of applications from materials science, chemical engineering, robotics, and energy systems.
Dr. Faruque Hasan is a full professor of Chemical Engineering and currently holds the Margaret '85 and Graham Bacon '85 Engineering Excellence Professorship at Texas A&M University. He also serves as an Associate Director of the Texas A&M Energy Institute. He received his B.Sc. in Chemical Engineering from Bangladesh University of Engineering & Technology in 2005 and a Ph.D. from National University of Singapore in 2010. After a postdoctoral fellowship at Princeton University, he joined Texas A&M University in 2014. His research interests include scientific machine learning and mixed-integer nonlinear optimization with applications to hybrid modeling, computer-aided process intensification, integrated molecular and process design, and multiscale energy systems engineering. Professor Hasan is the recipient of an NSF CAREER award, Outstanding Young Researcher Award from the AIChE Computing and Systems Technology (CAST) Division, I&ECR 2019 Class of Influential Researchers, and Best Paper Awards from Computers & Chemical Engineering and Journal of Global Optimization. He has co-authored over 150 peer-reviewed publications and advised over 25 PhD and MS students. He currently serves as the Second Vice Chair of the AIChE CAST Division and will assume the role of Division Chair in 2028.