Sparse Neural Approximations for Bilevel Adversarial Problems in Power Grids

TL;DR

This paper introduces a scalable, neural network-based approach for solving complex bilevel adversarial problems in power grids, enabling fast and accurate contingency analysis by leveraging physics constraints and sparse neural architectures.

Contribution

It develops a novel neural surrogate model embedded in a physics-constrained framework for efficient, scalable bilevel power grid analysis, improving over traditional mixed-integer methods.

Findings

Achieves an average optimality gap of 5.8% on IEEE test systems.

Maintains computation times under one minute for large-scale problems.

Scales linearly with system size due to sparse neural network architecture.

Abstract

The adversarial worst-case load shedding (AWLS) problem is pivotal for identifying critical contingencies under line outages. It is naturally cast as a bilevel program: the upper level simulates an attacker determining worst-case line failures, and the lower level corresponds to the defender's generator redispatch operations. Conventional techniques using optimality conditions render the bilevel, mixed-integer formulation computationally prohibitive due to the combinatorial number of topologies and the nonconvexity of AC power flow constraints. To address these challenges, we develop a novel single-level optimal value-function (OVF) reformulation and further leverage a data-driven neural network (NN) surrogate of the follower's optimal value. To ensure physical realizability, we embed the trained surrogate in a physics-constrained NN (PCNN) formulation that couples the OVF inequality with (relaxed) AC feasibility, yielding a mixed-integer convex model amenable to off-the-shelf solvers. To achieve scalability, we learn a sparse, area-partitioned NN via spectral clustering; the resulting block-sparse architecture scales essentially linearly with system size while preserving accuracy. Notably, our approach produces near-optimal worst-case failures and generalizes across loading conditions and unseen topologies, enabling rapid online recomputation. Numerical experiments on the IEEE 14- and 118-bus systems demonstrate the method's scalability and solution quality for large-scale contingency analysis, with an average optimality gap of 5.8% compared to conventional methods, while maintaining computation times under one minute.