ML for power systems at FAU Erlangen-Nürnberg: open protection benchmarks and peak-aware load forecasting.
4
papers here
4
reviewed by Evolve
3
first-author
0
corresponding
—
works · OpenAlex
2
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-15) and cover the author's entire public record, not just the papers listed here. Reproducibility signal comes from Evolve's own reviews.
Research profile
Output—
Not yet scored: this snapshot provides no computed output metric. The axis would measure total and per-year research production — relevant given 25 recent works — against field norms.
Impact—
No impact score: citation-based influence is not yet available in this snapshot; h-index 2 and i10-index 1 are too early to rank. The axis would measure citation uptake relative to field and career stage.
Focus0/100
Exceptional topical concentration (score 100): OpenAlex topic counts are dominated by power systems fault detection (18 works), followed by electrical fault detection and protection (7), islanding detection (6), power system optimization and stability (5), and smart grid security and resilience (4). A single field label — electric power systems (smart grids) / applied machine learning — confirms one coherent line: ML applied to power-grid monitoring and protection rather than dispersed interests.
Collaboration62/100
Collaboration score 48 with 10 distinct recent co-authors. The pattern is team-based within FAU Erlangen-Nürnberg: recurring partners include Christian Bergler, Siming Bayer, Souhardya Chattopadhyay, Antonia Schoening, Jessica Deuschel, Bitan Bhattacharyya, Andreas Maier, Georg Kordowich, and Johann Jäger. Oelhaf leads as first author on the benchmark paper but sits mid-author-list on the forecasting work, suggesting contribution across roles within a stable local group rather than a wide external network.
Activity94/100
Very high recent activity (score 95): OpenAlex countsByYear show 7 works in 2025 rising to 18 in 2026, and 25 works over the last three years — an unusually steep ramp for an early-career profile. 18 of the 25 recent works are indexed in 2026, matching the concentration of the two Evolve-reviewed arXiv papers in the same year.
Reproducibility69/100
Moderate-to-good reproducibility signal (score 55, from 1 evaluated paper). Evolve rates EvEMTBench's reproducibility readiness as high — an open, versioned benchmark with rigorous protocol — though its data are fully synthetic; the forecasting paper's readiness is medium. One paper is a thin base, so this score should be read as provisional until more outputs are assessed.
Research summary
Julian Oelhaf is an early-career researcher at Friedrich-Alexander-Universität Erlangen-Nürnberg working at the intersection of electric power systems and applied machine learning. OpenAlex records 25 works over the last three years, with annual output rising from 7 indexed works in 2025 to 18 in 2026 — a ramp reflected in an activity score of 95. Citation indicators remain modest (h-index 2, i10-index 1), so the output and impact dimensions are not yet scored. The topical footprint is sharply concentrated (focus 100), led by power systems fault detection (18 works), electrical fault detection and protection (7), islanding detection (6), power system optimization and stability (5), and smart grid security and resilience (4). Two Evolve-reviewed papers anchor the profile: EvEMTBench (2026), an open, versioned machine-learning benchmark for power system protection spanning 4 grids and 24 tasks, where Oelhaf is first author; and a peak-aware short-term load forecasting study across distribution grid aggregation levels (2026), evaluating time-series foundation models against gradient boosting. Evolve rates the benchmark's reproducibility readiness high despite fully synthetic data, while a single-paper reproducibility mean of 55 keeps that axis provisional. Collaboration (score 48) is compact — 10 distinct recent co-authors including Christian Bergler, Siming Bayer, Andreas Maier, and Johann Jäger.
Collaboration pattern
Oelhaf's collaboration profile is anchored at FAU Erlangen-Nürnberg with 10 distinct recent co-authors (collaboration score 48). Recurring names include Christian Bergler, Siming Bayer, Souhardya Chattopadhyay, Antonia Schoening, Jessica Deuschel, Bitan Bhattacharyya, Andreas Maier, Georg Kordowich, and Johann Jäger. Authorship varies from first author on the EvEMTBench benchmark to a middle position on the load-forecasting paper, consistent with contributing across roles inside a stable local group; no external institutional pattern is visible in the provided data.
Research keywords
Highlights
First-author open ML benchmark for power system protection
Peak-aware short-term load forecasting across grid levels
Timeline
- 2025First indexed publication activity in OpenAlex: 7 works recorded for this year per countsByYear.
- 2026Publishes EvEMTBench as first author — an open, versioned ML benchmark for power system protection covering 4 grids and 24 tasks; annual indexed output reaches 18 works.
- 2026Co-authors the peak-aware short-term load forecasting study comparing Chronos-2 against gradient boosting across distribution grid aggregation levels.
Papers in the Evolve catalog
- first author
EvEMTBench: An Open Benchmark for Machine Learning in Power System Protection
- first author
Offline Reinforcement Learning for Distribution-Grid Protection
Peak-Aware Short-Term Load Forecasting Across Distribution Grid Aggregation Levels
- first author
A Systematic Evaluation of Machine Learning Methods for Fault Detection and Line Identification in Electrical Power Grids
Frequent co-authors
About this profile
This profile is AI-structured by Evolve.Pub from public bibliographic data (OpenAlex) and Evolve's own paper reviews. It is not written or endorsed by the researcher unless the profile is claimed. Radar dimensions are computed from real bibliometrics with fixed formulas — raw figures are shown alongside; missing dimensions mean insufficient data, never zero. Errors? Claim the profile to correct or hide it.
- Bibliometric indicators (h-index, i10-index, countsByYear, topics, co-author counts) come from OpenAlex only and reflect a very early career stage; citation figures likely lag actual output.
- The output and impact dimensions are not scored in this snapshot due to insufficient underlying data.
- The reproducibility score (55) rests on a single evaluated paper and should be treated as provisional.
- Identity is name-resolved from public bibliographic data (OpenAlex, linked to ORCID 0009-0008-8204-589X); only 2 papers have been reviewed in depth by Evolve.