FSNet: Feasibility-Seeking Neural Network for Constrained Optimization with Guarantees

Priya Donti – MIT

Efficiently solving constrained optimization problems is crucial for numerous real-world applications, yet traditional solvers are often computationally prohibitive for real-time use. Machine learning-based approaches have emerged as a promising alternative to provide fast end-to-end surrogates for optimization solvers, but such approaches often struggle to strictly enforce constraints, leading to infeasible solutions in practice. To address this, we propose the Feasibility-Seeking Neural Network (FSNet), which integrates a feasibility-seeking step directly into its solution procedure to ensure constraint satisfaction. This feasibility-seeking step solves an unconstrained optimization problem that minimizes constraint violations in a differentiable manner, enabling end-to-end training and providing guarantees on feasibility and convergence. We show that this framework can deliver one to two orders of magnitude speedups for ACOPF on medium- and large-scale transmission and distribution test systems, while maintaining high-quality solutions with guaranteed feasibility.