Multi-Objective Reinforcement Learning for Power Grid Topology Control

• DOI: 10.48550/arxiv.2502.00040
• Publication Date: 2025-01-27
• Authors: Thomas Lautenbacher; Ali Rajaei; Davide Barbieri; Jan Viebahn; Jochen Cremer
• Keywords: Power grid

Transmission grid congestion increases as the electrification of various sectors requires transmitting more power. Topology control, through substation reconfiguration, can reduce but its potential remains under-exploited in operations. A challenge is modeling topology control problem to align well with objectives and constraints operators. Addressing this challenge, paper investigates application multi-objective reinforcement learning (MORL) integrate multiple conflicting for power control. We develop a MORL approach using deep optimistic linear support (DOL) proximal policy optimization (MOPPO) generate set Pareto-optimal policies that balance such minimizing line loading, topological deviation, switching frequency. Initial case studies show provide valuable insights into objective trade-offs improve Pareto front approximation compared random search baseline. The generated RL are 30% successful preventing failure under contingencies 20% effective when training budget reduced - common single policy.