Contact jjh348@cornell.edu for Zoom login information.
Machine Learning-Assisted Systems Optimization, Control, and Analytics
Abstract: Recent advances in machine learning, coupled with rapid growth in sensing capabilities and computing power, lead to increasing interest in learning-based techniques for systems optimization, control, and analytics. This presentation will introduce our recent theoretical, algorithmic, and computational results on machine learning-assisted stochastic optimization and control. The first part of the talk focuses on data-driven chance-constrained stochastic optimization. We will begin with the theoretical results of a posteriori probabilistic bounds of convex scenario programs with validation tests that optimize the use of out-of-sample data. Following the scenario program approach, we further present a deep learning-based distributionally robust chance-constrained optimization framework for economic dispatch in electric power systems considering uncertain wind power generation. In the second part of the presentation, we will discuss an online learning-based stochastic model predictive control (MPC) framework with data-driven ambiguity sets. We further establish theoretical guarantees on recursive feasibility and close-loop stability of the proposed stochastic MPC via a safe update scheme. We also develop a data-driven robust MPC framework for the irrigation application, by taking a learning-based approach and considering the dependence of uncertainty probability distributions to construct tailored uncertainty sets from historical forecast error data. The presentation will conclude with a novel deep learning model and quantum-assisted learning algorithm for efficient and effective fault detection and diagnosis in electric power systems and cyber-physical systems.
Bio: Fengqi You is the Roxanne E. and Michael J. Zak Professor at Cornell University, and is affiliated with the Graduate Fields of Chemical Engineering, Electrical and Computer Engineering, Systems Engineering, Operations Research and Information Engineering, Mechanical Engineering, Civil and Environmental Engineering, and Applied Mathematics. He also serves as Chair of Ph.D. Studies in Cornell Systems Engineering and Associate Director of Cornell Energy Systems Institute. His recent awards include NSF CAREER Award (2016), Computing and Systems Technology (CAST) Outstanding Young Researcher Award from AIChE (2018), Curtis W. McGraw Research Award from American Society for Engineering Education (2020), and American Automatic Control Council O. Hugo Schuck Award (2020), among others. He is currently an Editor of Computers & Chemical Engineering, an associate editor of AAAS journal Science Advances, and an associate editor of IEEE Transactions on Control Systems Technology. His research group website is: https://www.peese.org.
