Mingdong Lyu

Mingdong Lyu

Postdoc in Transporation @ NREL

About me

Postdoctoral Researcher at Mobility, Behavior, and Advanced Powertrains Group, National Renewable Energy Laboratory (NREL).

You can find my Curriculum Vitae here: Mingdong’s CV

Education

  • Ph.D. in Industrial Systems & Engineering, University of Southern California (USC), 2018-2023
  • Master of Science in Computer Science, University of Southern California (USC), 2020-2022
  • Master of Science in Electrical Engineering, Harbin Institute of Technology (HIT), 2016-2018
  • Bachelor of Science in Electrical Engineering, Harbin Institute of Technology (HIT), 2012-2016

Research Interests

  • Vehicle Routing Problem with uncertain travel time
  • Artificial Intelligence for Transportation
  • Large Scale Optimization
  • Graphic Neural Network
  • Statistical Modeling

Projects

  • Proximal Policy Optimization (PPO) Reinforcement Learning for Vehicle Routing with Routee Compass

    Applied PPO algorithms for vehicle routing with Routee Compass, optimizing cost efficiency by balancing time, distance, and energy. Achieved a 17% average reduction in operational costs by implementing these optimal routing policies.

  • Hierarchical Optimization Method (HOME) for eVTOL Network Design

    Initiated the design of a Hierarchical Optimization Method (HOME) for eVTOL network planning, combining integer linear programming with parallel computing. This method improved computational speed, reducing processing times from hours to minutes for complex network designs, and was validated through comparison with traditional methods in peer-reviewed forums.

  • Multi-model Route Energy Prediction under Uncertainty with Routee Powertrain Designed and deployed a suite of algorithms to forecast energy costs under uncertainty, supporting decisions for over 1.4 million road links. This suite, consisting of Random Forest, XGBoost regression, and MLP Neural Network, enabled Routee Powertrain to offer predictions with an 11% lower error margin, greatly aiding in route planning and energy management.
  • Self-attention Graph Convolution Network Model for International Air Passenger Flow Created a Self-Attention Graph Convolutional Network (SAGCN) to forecast international air passenger flows, employing Graph Convolution Network for node embedding with transformers for self-attention mechanisms. Demonstrated a 14% improvement in prediction accuracy over traditional deep gravity models.
  • Dynamic Multimodal Route Planner for Large-Scale Mobility Analysis Collaboratively developed a dynamic multimodal routing algorithm using A* search to integrate various travel modes, dynamically adjust travel times, and employ parallel AWS for performance improvement. This innovative approach reduced the computation time from three months to just 48 hours for analyzing the entire United States, showcasing a substantial leap in algorithmic efficiency and large-scale mobility analysis capabilities.
  • Integrated Modeling and Healthcare Resource Optimization in Pandemics Developed a compartmental model and dynamic optimization strategy for pandemic resource allocation, utilizing a novel Euler simplification for non-linear differential equations. This model guided vaccine distribution strategies, reducing case number by around 220,010 and fatalities by around 6,300.

Publication

  • Mingyi He, Mingdong Lyu et al., “A Hierarchical Optimization MEthod (HOME) for eVTOL Network Design”, Computer-Aided Civil and Infrastructure Engineering, 2023 (Under review)
  • Mingyi He, Mingdong Lyu et al., “GravAttn: A spatially-transferable gravity model for trip distribution based on Graph Attention Network and self-attention”, Transportation Research Board, 2023
  • Mingdong Lyu, R. Hall et al., Dynamic Vaccine Allocation for Control of Human Transmissible Disease, Health Care Management Science, 2023 (Under review)
  • Mingdong Lyu, R.Hall, A.Moore, “Modeling Reported Covid19 with Time Varying Case Fatality and Transmission Rates”, Computational and Mathematical Methods in Medicine, 2023 (Under review)
  • A. Moore, Mingdong Lyu, R. Hall, “Tracking Covid-19 Cases and Deaths in the United States Distribution of Events by Day of Pandemic”, Statistical Methods in Medical Research, 2021
  • Decker Nathan, Mingdong Lyu, Yuanxiang Wang, Qiang Huang. “Geometric Accuracy Prediction and Improvement for Additive Manufacturing Using Triangular Mesh Shape Data.” Journal of Manufacturing Sci. and Eng., June 2021
  • Q. Huang, Y. Wang, Mingdong Lyu, and W. Lin, “Shape Deviation Generator—A Convolution Framework for Learning and Predicting 3-D Printing Shape Accuracy”, IEEE Trans. Autom. Sci. Eng., January 2020

Awards

  • Health Systems Science and Innovation Student Innovators Fellowship 2022
  • USC Viterbi Ph.D. Fellowship in Industrial & Systems Engineering 2018~2019
  • Outstanding Graduate of Harbin Institute of Technology 2018
  • First Prize Academic Scholarship of Harbin Institute of Technology 2016~2017
  • First Prize Scholarship Sponsored by Infineon Technology 2016~2017
  • First Prize Scholarship for Graduate Student of Harbin Institute of Technology 2015~2016
  • Outstanding Graduate Thesis of Harbin Institute of Technology 2016
  • Meritorious Winner of Mathematical Contest in Modeling 2015