Academic Integrity Review of 'Deep-Quantile-Regression-Based Surrogate Model for Joint Chance-Constrained Optimal Power Flow With Renewable Generation' (DOI: 10.1109/TSTE.2022.3223764)

Academic fraud report · Geng Detector

Summary

Verdict: Cleared (no substantive academic misconduct detected); one minor typographical defect noted. The reviewers examined logical consistency, submission timeline, methodology, mathematical derivations, and open-source declarations using only the plain-text extracted from the PDF. The manuscript timeline (submitted 1 April 2022, revised three times, accepted November 2022, published January 2023) is consistent with normal high-journal peer review. Tools cited (XGBoost 2016, Pandapower 2018, PyTorch, CVXPY) and references predate submission. The authors released code and data to a public GitHub repository and used the real Kaggle HI-SEAS solar irradiance dataset, which greatly reduces the risk of fabricated experiments. Propositions 1 and 2 use standard Pinball-loss and MSE derivations with no sign of contrived mathematics. The sole defect is a duplicated subfigure label in the Figure 6 caption: '(d)' is printed twice and the sequence skips from (d) to (e), almost certainly a copy-paste error. No image-level, statistical, or data-fabrication analyses could be performed because raw figures and numerical tables were not provided.

Verdict

Cleared. No evidence of academic fraud. One minor formatting/labeling typo identified in Figure 6 caption.

Key findings

Figure 6 caption contains a duplicate subfigure label: '(d)' appears twice and the sequence jumps from (d) to (e), consistent with a copy-paste or typesetting slip rather than misconduct.
Submission and revision timeline (received 1 Apr 2022; revisions Jul/Oct/Nov 2022; accepted Nov 2022; published Jan 2023) is fully consistent with normal peer review at a top venue.
All cited software (XGBoost, Pandapower, PyTorch, CVXPY) and reference [41] (early-access article from March 2022) predate submission; no temporal anomalies.
Authors declare open availability of samples in a public GitHub repository (Ref. [55]: https://github.com/lelouchsola/DeepQuantileRegressionOPF) and use a real Kaggle dataset (HI-SEAS), supporting transparency.
Mathematical content in Appendix A/B/C (Pinball-loss expectation derivation in Proposition 1, MSE mean-minimizer argument in Proposition 2) is logically coherent and uses standard results.

Evidence highlights

DOI: 10.1109/TSTE.2022.3223764
Manuscript timestamps: received 1 April 2022; revised July 2022, October 2022, November 2022; accepted November 2022; published January 2023 (IEEE Transactions on Sustainable Energy, Vol. 14, No. 1).
Figure 6 caption excerpt (Page 665): "Comparisons of actual and predicted values of (a) 80% quantile... (b) expected power loss... (c) 80% quantile... (d) expected power loss in Case 2 (Beta uncertainties), (d) 80% quantile of constraint violation in Case 3 (Weibull uncertainties), (e) expected power loss in Case 3."
Open-data/code statement (Page 664, Section IV.A): "All these samples have been uploaded to [55]" with Ref. [55] linking to the GitHub repository above.
HI-SEAS Kaggle dataset used as Case 4 input.

Notes

Scope limitation: only plain-text PDF content was available. Pixel-level image forensics (Western blot splicing, microscopy duplication, flow cytometry cloning), grayscale analysis, and last-digit statistical tests could not be performed. Findings are limited to textual, logical, methodological, and bibliographic consistency.
The Figure 6 caption defect is classified as a careless typesetting error; it does not affect scientific validity or reproducibility.
Confidence in cleared verdict: moderate-to-high for text-based checks; low-to-none for visual-based checks. A final determination of misconduct should be made by qualified institutional investigation.
No contact with the authors is required; data and code are already publicly released.