Industrial AI researcher: reinforcement learning, digital twins and MLOps for manufacturing control and maintenance
3
papers here
3
reviewed by Evolve
0
first-author
0
corresponding
—
works · OpenAlex
—
h-index · OpenAlex
Bibliometric figures come from OpenAlex (2026-10-13) 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—
No total publication count or productivity data is available yet; this axis would measure the volume and consistency of output over time. Current evidence covers only the three arXiv papers in the Evolve catalog.
Impact—
Citation counts, h-index and downstream influence are not yet available; this axis would measure how often and how widely the work is cited and adopted. No impact score is estimated here.
Focus0/100
Focus scores 0 because the three catalog papers split evenly across three fields — industrial MLOps for manufacturing (0.33), industrial AI/predictive maintenance (0.33), and reinforcement learning for industrial control (0.33) — with no dominant line. Read charitably, this is a coherent industrial-AI cluster around manufacturing control and maintenance rather than scatter; but with only three papers the concentration signal is unstable.
Collaboration48/100
Collaboration scores 41 from 7 distinct coauthors across the three catalog papers; all papers list Gu in a middle author position, none as corresponding author. Recurring partners include Zhengyang, Thomas Cook, Joseph E. Hernandez, Chris Couch, John Burtenshaw, Sean Scott and Fredaljohn Rohrbaugh. The moderate score reflects a small, consistent team rather than a wide network.
Activity—
No per-year productivity or publication cadence data is available; this axis would measure recent output rate and momentum. The only dated evidence is three arXiv papers posted in 2026, which is too small a window to score.
Reproducibility32/100
Reproducibility scores 32 (mean across 3 reviewed papers). The digital-twin MLOps paper and the PID+TD3 hybrid are rated low readiness — internal-only evaluation, no code or data, single-episode results, no significance tests — while the predictive-maintenance benchmark rates medium (700+ runs, but synthetic linear targets and missing linear baselines).
Research summary
This profile covers Gu, name-resolved from public bibliographic data on a surname alone; no ORCID, institution or country is on record, so identity confidence is limited. The evidence base is three arXiv preprints, all dated 2026, in which Gu appears as a middle author and never as corresponding author. The work forms a coherent applied cluster around industrial AI. One paper describes autonomous model lifecycle management for digital-twin-based manufacturing control — a deployed automotive MLOps system reporting Cpk gains of 28–45% over uncontrolled operation, though evaluated internally only. Another proposes a hybrid PID controller whose TD3-discovered Gain/Shift adjustments match the best deep-RL agent on the Industrial Benchmark (-203.46 versus -204.92), based on single-episode evaluation. The third benchmarks hybrid deep learning architectures for Industry 4.0 predictive maintenance over 700+ runs, claiming LSTM-Transformer superiority in noise robustness. Fields split evenly three ways — manufacturing MLOps, industrial AI/predictive maintenance, and reinforcement learning for industrial control (0.33 each) — producing a focus score of 0. Collaboration scores 41 from 7 distinct coauthors; reproducibility scores 32, with two papers rated low readiness and one medium. Output, impact and activity metrics are not yet available.
Collaboration pattern
Gu collaborates within a small, stable team: 7 distinct coauthors across the three 2026 arXiv papers, yielding a collaboration score of 41. Recurring names include Zhengyang, Thomas Cook, Joseph E. Hernandez, Chris Couch, John Burtenshaw, Sean Scott and Fredaljohn Rohrbaugh. On every paper Gu sits in a middle author position and is never the corresponding author, suggesting a contributor role within a group rather than leading an independent collaboration network.
Research keywords
Highlights
Deployed digital-twin MLOps with Cpk gains of 28–45%
Hybrid PID + TD3 control matches the best DRL agent
700+ run benchmark of hybrid architectures for predictive maintenance
Timeline
- 2026Releases a benchmark of hybrid deep learning architectures (LSTM, Transformer) for Industry 4.0 predictive maintenance, based on 700+ runs.
- 2026Publishes a hybrid PID control approach augmented by TD3-discovered Gain/Shift adjustments, evaluated on the Industrial Benchmark.
- 2026Posts an autonomous model lifecycle management system for digital-twin-based manufacturing control, a deployed automotive MLOps system reporting Cpk gains of 28–45%.
Papers in the Evolve catalog
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.
- Identity is resolved from the surname 'Gu' only; no ORCID, institution or country is on record, so the profile may mix in or miss work by same-named authors.
- Coverage is limited to the three Evolve-reviewed arXiv preprints (all dated 2026); citation counts, h-index and venue history are not available.
- Radar scores (collaboration 41, focus 0, reproducibility 32) are computed from only 3 papers and 7 distinct coauthors — early-stage signals, not stable estimates.
- No peer-reviewed venue, funding or award information is included; absence of data here is not evidence of absence.