The rapid evolution of machine learning algorithms has ushered in a transformative era for smart infrastructure development. This presentation unveils the potential of machine learning through two compelling case studies, illustrating how it can drive cost optimization, enhance efficiency, and elevate the reliability of infrastructure systems.

The first case study delves into the realm of reinforcement learning, showcasing its application in orchestrating an electric vehicle (EV) fleet for ride-hailing services. The goals of operating the EV fleet are to minimize customer waiting time, electricity cost, and operational costs of the vehicles. A novel framework characterized by decentralized learning and centralized decision making is proposed to solve the EV fleet dispatch problem. The decentralized learning process allows the individual vehicles to share their operating experiences and deep neural network model for state-value function estimation, which mitigates the curse of dimensionality of state and action domains. The centralized decision-making framework converts the vehicle fleet coordination problem into a linear assignment problem, which has only polynomial time complexity. Numerical studies validate the proposed approach, demonstrating superior societal cost reduction over benchmark algorithms.

The second case study presents a novel power system event identification framework, harnessing the capabilities of deep learning. Specifically, a deep convolutional neural network (CNN) based approach is developed to identify power system events by leveraging real-world measurements from hundreds of phasor measurement units (PMUs) and labels from thousands of events. Enriching the baseline model, two innovative designs emerge as keystones of improvement. First, a graph signal processing-based PMU sorting algorithm is proposed to improve the learning efficiency of CNNs. Second, a pioneering information loading-based regularization technique is adopted to strike the right balance between memorization and generalization for the neural network. Numerical results based on real-world dataset from the Eastern Interconnection of the U.S power transmission grid show that the proposed approach achieves remarkably accurate event identification and classification results.

Dr. Jie Shi is a senior data scientist at Sams AI Lab within Walmart Global Tech. Prior to joining Walmart, he was a postdoctoral researcher with Systems Engineering at Cornell University. He received a B.S. degree in Automation from Shenyang University of Technology in 2012, and an M.S. degree in Control Theory & Control Engineering from Southeast University in 2015. He received his Ph.D. degree in Electrical Engineering from University of California, Riverside, CA in 2021. He is the recipient of IEEE Power and Energy Society (PES) Technical Committee Prize Paper Award and multiple best paper awards from international conferences organized by IEEE PES. His research interests include smart infrastructure, intelligent energy systems, and machine learning.