LEAP nets for power grid perturbations
Source
- Raw Markdown: leap-net-power-grid-perturbations-2019
- Rendered / retrieved PDF: paper_leap-net-power-grid-perturbations-2019.pdf
- External source: https://arxiv.org/abs/1908.08314
- Official L2RPN reference list: https://l2rpn.chalearn.org/papers-references
Publication And Credibility
- Paper date: 2019-08-22.
- Venue/status: ESANN 2019 lineage; arXiv preprint available.
- Credibility: Credible arXiv source by the LEAP/RTE/ChaLearn lineage. Older than one year; use as architecture-lineage context.
Core Claim
The paper introduces LEAP nets for generalizing across power-grid structural perturbations.
L2RPN / Grid2Op Notes
LEAP nets make structural changes visible to the model instead of treating topology changes as hidden distribution shift.
Action-Time-Series Notes
This source is useful when Grid2Op is treated as an action-conditioned graph time-series environment:
power-grid observations + topology / redispatch / storage control input + scenario context
-> next grid observations + safety/cost outcomeThe terminology distinction matters. Topology changes, redispatching, curtailment, and storage commands are actions or control inputs when an agent chooses them. Line failures, maintenance outages, weather-driven renewable shifts, and demand variation are events or exogenous variables unless they are deliberately controlled by the experimenter.
In this paper’s synthetic setup, tau is an explicit topology variable. In the real Toulouse records, exact grid topologies were not recorded and line outages were only a surrogate, so the real-data result is observational topology-surrogate evidence rather than clean logged-action data.
Foundation TSFM Relevance
| Agenda slot | Verdict | Evidence | Missing pieces |
|---|---|---|---|
| Causal structure, counterfactuals, and control | adjacent | Relevant to TSFM context interfaces because topology and structural actionable variables must be first-class inputs. | It focuses on prediction/generalization under perturbations, not sequential action selection, and the real-grid records do not contain exact topology intervention logs. |
| Context interface: topology and channel context | partially closes | Power-grid state is naturally graph-structured and tied to physical assets, limits, and scenario metadata. | Needs a reusable schema that a general TSFM can consume across grids and non-grid operational systems. |
| Benchmark level | adjacent | L2RPN/Grid2Op provides simulator-backed trajectories with explicit controls and outcomes. | TSFM-ready comparisons require pinned environment versions, action sets, reward definitions, and train/test scenario splits. |
Tension: synthetic experiments expose tau as a structural variable, but real RTE records do not contain exact topology interventions. Do not treat the real-data result as a clean action-conditioned transition dataset.