# Interpretable Policy Distillation for Power Grid Topology Control

arXiv: https://arxiv.org/abs/2606.00561 · HTML: https://mlsanity.com/paper/2606.00561/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in RL for energy, networks and operations, one of 15 directions in Deep reinforcement learning methods.
- Builds on 9 of its 18 references that are in the map.
- First posted on arXiv 30 May 2026.

## Placed in

- Deep reinforcement learning methods / [RL for energy, networks and operations](https://mlsanity.com/deeprl/rl-for-energy-networks-and-operations/)

## Builds on (most cited first, 9 of 9)

- [Distilling the Knowledge in a Neural Network](https://mlsanity.com/paper/1503.02531/) (2015)
- [Verifiable Reinforcement Learning via Policy Extraction](https://mlsanity.com/paper/1805.08328/) (2018)
- [Action Set Based Policy Optimization for Safe Power Grid Management](https://mlsanity.com/paper/2106.15200/) (2021)
- [Reinforcement Learning for Electricity Network Operation](https://mlsanity.com/paper/2003.07339/) (2020)
- [HUGO -- Highlighting Unseen Grid Options: Combining Deep Reinforcement Learning with a Heuristic Target Topology Approach](https://mlsanity.com/paper/2405.00629/) (2024)
- [RL2Grid: Benchmarking Reinforcement Learning in Power Grid Operations](https://mlsanity.com/paper/2503.23101/) (2025)
- [Graph Reinforcement Learning for Power Grids: A Comprehensive Survey](https://mlsanity.com/paper/2407.04522/) (2024)
- [Optimizing Interpretable Decision Tree Policies for Reinforcement Learning](https://mlsanity.com/paper/2408.11632/) (2024)
- [Graph-Enhanced Model-Free Reinforcement Learning Agents for Efficient Power Grid Topological Control](https://mlsanity.com/paper/2503.20688/) (2025)

## Abstract (arXiv)

Deep reinforcement learning (RL) offers a promising route to real-time power grid operation, yet large neural policies are costly to evaluate, hard to deploy on constrained hardware, and opaque to operators. We ask whether a Proximal Policy Optimization (PPO) agent for grid topology control can be compressed into compact tree-based surrogates without losing operational performance. A PPO teacher is trained on Grid2Op's standard 14-bus environment with a stability-oriented reward, using stress-focused data collection on critical, high-loading states. The policy is then distilled into a decision tree and a random forest. Across held-out validation episodes, both surrogates exceed the teacher in mean reward and survival length at a fraction of the inference cost. The decision tree shows high exact-action agreement with the PPO argmax and near-complete agreement within its top-ranked actions, while remaining small enough to be inspected directly. Feature-importance analysis reveals a representational shift: the PPO policy relies mainly on line-loading signals, while the distilled tree is driven primarily by bus-topology variables. These results suggest that stress-focused distillation can convert a black-box neural controller into a lightweight, auditable rule-like surrogate suited for real-time deployment, while also surfacing risks tied to deterministic actions and topology-specific generalization.