Deep Constrained Learning

A variety of computationally challenging optimization problems in several engineering disciplines are solved repeatedly under different scenarios. In many cases, they would benefit from fast and accurate approximations, either to support real-time operations or large-scale simulation studies. These optimization problems arise in numerous contexts including in energy systems, mobility, resilience, and disaster management. These applications must capture physical laws such as Ohm’s law and Kirchhoff’s law in electrical power systems, the Weymouth equation in gas networks, flow constraints in transportation models, and the Navier-Stoke’s equations for shallow water in flood mitigation. Moreover, they often feature constraints that represent good engineering and operational practice to protect various devices.
This main direction of my research on Deep Learning for Optimization aims at exploring how to leverage deep learning models to aid the resolution of constrained optimization problems and, in particular, optimization problems with hard physical and engineering constraints.


Students and Collaborators


Collaborative Research: RI: Small: Deep Constrained Learning for Power Systems
National Science Foundation (2020 – 2023).

Recent Publications