Geng Integrity Report: Deep Quantile Regression Surrogate Model for Joint Chance-Constrained Optimal Power Flow With Renewable Generation

Academic fraud report · Geng Detector

Summary

Verdict: CLEAN (no integrity issues found). The IEEE Transactions on Sustainable Energy paper (DOI: 10.1109/TSTE.2022.3223764) by Ge Chen, Hongcai Zhang, Hongxun Hui, and Yonghua Song was screened across multiple dimensions including image reuse, data fabrication, statistical anomalies, and timeline plausibility, and no irregularities were identified. The paper relies on engineering-style figures (scatter plots, bar charts, line plots, architecture diagrams) rather than biological images, eliminating typical image-manipulation risks. Reported quantitative results show expected imperfections, such as elevated quantile-prediction error near Quantile = -0.2 and SOCP-relaxation violation probability approaching 100% under high renewable penetration, which is consistent with the underlying physics rather than curated data. No hypothesis-testing statistics are used, so p-hacking is not applicable. The submission-to-acceptance timeline (~7 months, three revisions), the Intel 8700 CPU (released 2018), and the 2022.08.01 HI-SEAS dataset access all align with the 2022/2023 publication dates. Notably, the authors openly released code and samples at github.com/lelouchsola/DeepQuantileRegressionOPF, providing a strong reproducibility safeguard. Limitations: this is a heuristic review, not an institutional investigation.

Verdict

CLEAN — No indicators of academic fraud detected.

Key findings

No image reuse, flipping, cropping, splicing, or pixel-level duplication detected across Figures 3, 4, 6, and 7–15. The figure types (architectural diagrams, scatter plots, bar/line plots) are not the high-risk categories typically associated with image manipulation.
Quantitative results contain honest imperfections: quantile-regression error is visibly larger near Quantile = −0.2 in Figures 6(a), (c), (e), and the SOCP relaxation (B2) reports violation probabilities approaching 100% under high renewable penetration in Figures 7–13 — consistent with physics rather than data curation.
No traditional inferential statistics (p-values, ANOVA) are used; validation relies on engineering metrics (optimality, feasibility) and Monte Carlo violation-probability simulations.
Timeline is plausible: received 2022.04.01, accepted 2022.11.17, spanning ~7 months with three revisions, consistent with IEEE Transactions peer-review norms.
No hardware or data anachronism: Intel 8700 (3.20 GHz) CPU was released in 2018; HI-SEAS solar irradiance data accessed on 2022.08.01, both prior to acceptance.
Authors openly released code and data at github.com/lelouchsola/DeepQuantileRegressionOPF (reference [55]), providing a strong reproducibility anchor.

Evidence highlights

DOI: 10.1109/TSTE.2022.3223764
Journal: IEEE Transactions on Sustainable Energy, Vol. 14, No. 1, January 2023
Open-source artifact: Section IV-A states "All these samples have been uploaded to [55]" with the GitHub link above.
Timeline evidence: Received 2022.04.01; Accepted 2022.11.17.
Hardware evidence: Intel(R) 8700 @ 3.20 GHz CPU (released 2018).
Data-access evidence: HI-SEAS dataset accessed 2022.08.01.

Notes

This report is an AI-assisted heuristic screening and does not constitute a formal institutional investigation; false negatives remain possible.
All figures mentioned were visually inspected only; no forensic-level pixel analysis was performed.
The high transparency of the authors (open code + data) materially reduces reproducibility risk and is treated as a positive integrity signal rather than as proof against past misconduct.