Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Transferable Graph Learning for Transmission Congestion Management via Busbar Splitting

Ali Rajaei, Peter Palensky, Jochen L. Cremer

TL;DR

This work tackles congestion management via busbar splitting by formulating a tractable MILP using a linearized AC power-flow model (AC-CM-NTO) and introducing a transferable, edge-aware GNN (GNN-NTO) to rapidly generate high-quality busbar-splitting candidates. The approach leverages locality through a proximity filter and two learning tasks (classification and regression) to predict effective actions, while a heterogeneous, fully-local GNN architecture ensures transferability across grid sizes and different systems. Case studies on IEEE and GOC networks show up to $10^4$× speed-ups for large-scale AC PF cases and a small optimality gap (e.g., around $2.3 ext{–}6 ext{%}$), with strong generalization to topology changes and cross-system transferability (including TL and CD configurations). The results indicate that near-real-time NTO for large-scale grids is feasible and that cross-system, topology-agnostic learning can significantly reduce data requirements and computation while maintaining feasibility. Overall, the paper demonstrates a practical pathway to scalable, transferable topology optimization in modern power systems, with potential for sequential topology reconfigurations and zero-shot transfer in future work.

Abstract

Network topology optimization (NTO) via busbar splitting can mitigate transmission grid congestion and reduce redispatch costs. However, solving this mixed-integer non-linear problem for large-scale systems in near-real-time is currently intractable with existing solvers. Machine learning (ML) approaches have emerged as a promising alternative, but they have limited generalization to unseen topologies, varying operating conditions, and different systems, which limits their practical applicability. This paper formulates NTO for congestion management problem considering linearized AC PF, and proposes a graph neural network (GNN)-accelerated approach. We develop a heterogeneous edge-aware message passing NN to predict effective busbar splitting actions as candidate NTO solutions. The proposed GNN captures local flow patterns, achieves generalization to unseen topology changes, and improves transferability across systems. Case studies show up to 4 orders-of-magnitude speed-up, delivering AC-feasible solutions within one minute and a 2.3% optimality gap on the GOC 2000-bus system. These results demonstrate a significant step toward near-real-time NTO for large-scale systems with topology and cross-system generalization.

Transferable Graph Learning for Transmission Congestion Management via Busbar Splitting

TL;DR

This work tackles congestion management via busbar splitting by formulating a tractable MILP using a linearized AC power-flow model (AC-CM-NTO) and introducing a transferable, edge-aware GNN (GNN-NTO) to rapidly generate high-quality busbar-splitting candidates. The approach leverages locality through a proximity filter and two learning tasks (classification and regression) to predict effective actions, while a heterogeneous, fully-local GNN architecture ensures transferability across grid sizes and different systems. Case studies on IEEE and GOC networks show up to × speed-ups for large-scale AC PF cases and a small optimality gap (e.g., around ), with strong generalization to topology changes and cross-system transferability (including TL and CD configurations). The results indicate that near-real-time NTO for large-scale grids is feasible and that cross-system, topology-agnostic learning can significantly reduce data requirements and computation while maintaining feasibility. Overall, the paper demonstrates a practical pathway to scalable, transferable topology optimization in modern power systems, with potential for sequential topology reconfigurations and zero-shot transfer in future work.

Abstract

Network topology optimization (NTO) via busbar splitting can mitigate transmission grid congestion and reduce redispatch costs. However, solving this mixed-integer non-linear problem for large-scale systems in near-real-time is currently intractable with existing solvers. Machine learning (ML) approaches have emerged as a promising alternative, but they have limited generalization to unseen topologies, varying operating conditions, and different systems, which limits their practical applicability. This paper formulates NTO for congestion management problem considering linearized AC PF, and proposes a graph neural network (GNN)-accelerated approach. We develop a heterogeneous edge-aware message passing NN to predict effective busbar splitting actions as candidate NTO solutions. The proposed GNN captures local flow patterns, achieves generalization to unseen topology changes, and improves transferability across systems. Case studies show up to 4 orders-of-magnitude speed-up, delivering AC-feasible solutions within one minute and a 2.3% optimality gap on the GOC 2000-bus system. These results demonstrate a significant step toward near-real-time NTO for large-scale systems with topology and cross-system generalization.
Paper Structure (15 sections, 7 equations, 4 figures, 4 tables)

This paper contains 15 sections, 7 equations, 4 figures, 4 tables.

Table of Contents

  1. Introduction
  2. Congestion Management via Busbar Splitting
  3. Transferable Graph Learning Approach
  4. GNN-accelerated NTO approach
  5. Learning Task
  6. Transferable Heterogeneous GNN Architecture
  7. Edge-Aware Message Passing
  8. Heterogeneous GNN
  9. Case Studies
  10. Setting and Test Networks
  11. Prediction Accuracy
  12. Generalization to Topology Changes
  13. Transferability across Systems
  14. Computational Efficiency
  15. Conclusion and Future Work

Figures (4)

  • Figure 1: Schematic of the substation model for AC power flow.
  • Figure 2: An example of congestion in the IEEE 118-bus system. Left: potential busbar splitting actions within 3 hops. Right: line flows impacted by the splitting action.
  • Figure 3: Workflow of the proposed GNN-accelerated NTO approach.
  • Figure 4: Schematic of the proposed encoder-GNN-decoder architecture.