AC Power Flow Contingency Analysis Using a Single Deep Neural Network

2026-09-25

AI Summary

This study addresses the high computational burden of traditional AC power flow contingency analysis and the limitation of existing machine learning methods that require per-contingency offline training, with costs scaling linearly. To overcome these challenges, this work proposes a unified framework leveraging a single deep neural network. Trained exclusively on base-case operating conditions, the framework estimates post-contingency grid states for any single-line outage. The core innovation lies in formulating post-contingency state prediction as a fixed-point iteration, deriving sufficient convergence conditions, and constructing a semidefinite programming formulation to provide theoretical certification. Experimental validation on the IEEE 118-bus system demonstrates the method’s tightness and generalizability, achieving highly accurate post-contingency state estimation within only a few iterations.

Abstract

Contingency analysis using the AC power flow (AC-PF) model is a critical tool for accurate grid security assessment, but its computational burden increases with the number of operating scenarios and outage configurations to evaluate. Recent ML-based approaches typically require outage-specific training data, leading to offline training costs that scale with the number of contingencies. This work proposes a framework that reuses a single ML model trained solely on basecase AC-PF data to estimate post-contingency operating states under arbitrary single-line outages. The proposed approach formulates post-contingency state prediction as a fixed-point iteration. If the ML model is a deep neural network (DNN), we derive sufficient conditions that guarantee convergence and develop semidefinite programming (SDP) formulations to certify these conditions for a given DNN. Numerical tests on the IEEE 118-bus system demonstrate that the proposed SDP formulations are tight, that the certified conditions hold for all tested contingencies, and that the resulting method produces accurate post-contingency state estimates within only a few iterations.

Problem

• Contingency analysis
• AC power flow
• Deep neural network
• Grid security assessment
• Computational burden

Innovation

• Contingency Analysis
• Deep Neural Network
• Fixed-Point Iteration
• Semidefinite Programming
• AC Power Flow