Image Integrity Report: "Optimizing flapping foil dynamics: A data-driven framework for motion optimization"

Academic fraud report · Geng Detector

Summary

This report assesses potential integrity concerns in a 2025 Ocean Engineering article (DOI: 10.1016/j.oceaneng.2025.122990) by Jinyu Li, Ruipeng Li, Jiaye Gong, Weicheng Cui, and Dixia Fan. The paper proposes a CFD-based framework combining PPO reinforcement learning, Gaussian Process Regression, and Transformer components to optimize flapping-foil motion. The overall verdict is "Questionable" rather than "Fabricated." Two issues were flagged: (1) multiple low-level typographical errors and encoding artifacts were confirmed in Algorithm 1 and the Conclusion (e.g., "perfermance," "squeues," "uur study," and mojibake entities such as &#3627408447;). These appear to be genuine proofreading or typesetting oversights rather than deliberate falsification. (2) Image-level analysis of Figures 2–7 could not be performed because raw high-resolution images were not available, so reuse of CFD vorticity fields cannot be ruled in or out. The numerical values in Table 1 are internally consistent with physical Pareto-front behavior. Confidence is moderate; the textual evidence is solid but the image dimension remains unresolved.

Verdict

Questionable (🟡). The article exhibits notable textual errors and encoding artifacts, but no direct evidence of data fabrication or figure reuse was identified. The verdict is constrained by the absence of pixel-level image analysis.

Key findings

Confirmed typographical errors in Algorithm 1: "perfermance" (for performance), "squeues" (for sequences).
Confirmed typo in the last paragraph of the Conclusion: "uur study" (for our study).
Presence of mojibake/HTML-entity artifacts (e.g., &#3627408447;) throughout the extracted text, likely arising from failed parsing of equations or special symbols.
The paper combines CFD with PPO reinforcement learning, Gaussian Process Regression, and Transformer-based components; Table 1 data are internally consistent with flapping-foil physics (thrust–efficiency Pareto trade-off).
Timeline is logically consistent: submitted June 2025, revised and accepted September 2025, with references as recent as 2024.
Image-integrity analysis of Figures 2–7 could not be completed because high-resolution originals were not available.

Evidence highlights

Direct textual evidence: "perfermance," "squeues," and "uur study" appear verbatim in Algorithm 1 and the Conclusion of the supplied PDF (1-s2.0-S0029801825026733-main.pdf).
Numeric evidence: Table 1 reports thrust coefficient and efficiency values for points 1–4 that obey the expected monotonic trend in which higher thrust corresponds to lower efficiency.
DOI: 10.1016/j.oceaneng.2025.122990 — confirms publication in Ocean Engineering, 2025.

Notes

The findings are limited to textual review; the absence of raw high-resolution figures means reuse of CFD vorticity plots cannot be excluded.
The detected errors are consistent with non-native English authorship or rushed typesetting rather than with intentional misconduct.
Recommended next steps: obtain original figure files and source CFD output to enable pixel-level duplication checks (e.g., FFT/ELA) on Figures 2–7; consider a PubPeer comment requesting raw data.