Kolmogorov–Arnold Network for Predicting CO2 Corrosion and Performance Comparison with Traditional Data-Driven Approaches

Dong, Zhenzhen; Zou, Lu; Xu, Yiming; Guo, Chenhong; Wen, Fenggang; Wang, Wei; Qi, Ji; Zhang, Min; Dong, Guoqing; Li, Weirong

doi:10.3390/pr13103174

This is an early access version, the complete PDF, HTML, and XML versions will be available soon.

Open AccessArticle

Kolmogorov–Arnold Network for Predicting CO₂ Corrosion and Performance Comparison with Traditional Data-Driven Approaches

by

Zhenzhen Dong

¹,

Lu Zou

¹,

Yiming Xu

¹,

Chenhong Guo

¹,

Fenggang Wen

²,

Wei Wang

¹,

Ji Qi

¹,

Min Zhang

³,

Guoqing Dong

⁴ and

Weirong Li

^1,*

¹

Petroleum Engineering College, Xi’an Shiyou University, No. 18, East Section of Dianzi 2nd Road, Yanta District, Xi’an 710065, China

²

Shaanxi Key Laboratory of Carbon Dioxide Sequestration and Enhanced Oil Recovery, No. 18, East Section of Dianzi 2nd Road, Yanta District 18 Xianning West Road, Xi’an 710065, China

³

Shaanxi Yanchang Petroleum (Group) Co. Ltd., No. 75, Keji 2nd Road, High-tech Zone, Xi'an 710075, China

⁴

Western Project Department, PetroChina Jidong Oilfield Company, Yulin 719000, China

^*

Author to whom correspondence should be addressed.

Processes 2025, 13(10), 3174; https://doi.org/10.3390/pr13103174

Submission received: 26 August 2025 / Revised: 19 September 2025 / Accepted: 26 September 2025 / Published: 6 October 2025

(This article belongs to the Section Chemical Processes and Systems)

Download Versions Notes

Abstract

Accurate prediction of CO₂ corrosion under dense-phase and supercritical conditions remains a critical challenge for oil and gas pipeline integrity management. While machine learning (ML) has been applied in this field, prevailing models like the Multilayer Perceptron (MLP) often struggle to capture the complex, non-linear interactions between multiple environmental parameters, limiting their predictive accuracy and robustness. To bridge this gap, this study innovatively introduces the Kolmogorov–Arnold Network (KAN) algorithm for CO₂ corrosion rate prediction. Utilizing a unique dataset of field-collected parameters (including dissolved O₂, H₂S, SO₂ concentrations, and water cut), we developed a KAN model and conducted systematic hyperparameter optimization. Our investigation revealed the optimal network configuration (3 layers, grid = 3) and, counterintuitively, that the steps parameter does not correlate positively with performance. Most significantly, comparative experiments demonstrated that the KAN model substantially outperforms traditional MLP, achieving superior prediction accuracy alongside faster computational speed and lower loss values. These findings not only provide a robust tool for precise corrosion prevention in oilfield operations but also highlight the potential of KAN as a novel, efficient, and highly accurate framework for tackling complex problems in materials degradation.

Keywords: CO₂ corrosion; hyperparameter optimization; Kolmogorov–Arnold Network; multilayer perceptron

Share and Cite

MDPI and ACS Style

Dong, Z.; Zou, L.; Xu, Y.; Guo, C.; Wen, F.; Wang, W.; Qi, J.; Zhang, M.; Dong, G.; Li, W. Kolmogorov–Arnold Network for Predicting CO₂ Corrosion and Performance Comparison with Traditional Data-Driven Approaches. Processes 2025, 13, 3174. https://doi.org/10.3390/pr13103174

AMA Style

Dong Z, Zou L, Xu Y, Guo C, Wen F, Wang W, Qi J, Zhang M, Dong G, Li W. Kolmogorov–Arnold Network for Predicting CO₂ Corrosion and Performance Comparison with Traditional Data-Driven Approaches. Processes. 2025; 13(10):3174. https://doi.org/10.3390/pr13103174

Chicago/Turabian Style

Dong, Zhenzhen, Lu Zou, Yiming Xu, Chenhong Guo, Fenggang Wen, Wei Wang, Ji Qi, Min Zhang, Guoqing Dong, and Weirong Li. 2025. "Kolmogorov–Arnold Network for Predicting CO₂ Corrosion and Performance Comparison with Traditional Data-Driven Approaches" Processes 13, no. 10: 3174. https://doi.org/10.3390/pr13103174

APA Style

Dong, Z., Zou, L., Xu, Y., Guo, C., Wen, F., Wang, W., Qi, J., Zhang, M., Dong, G., & Li, W. (2025). Kolmogorov–Arnold Network for Predicting CO₂ Corrosion and Performance Comparison with Traditional Data-Driven Approaches. Processes, 13(10), 3174. https://doi.org/10.3390/pr13103174

Note that from the first issue of 2016, this journal uses article numbers instead of page numbers. See further details here.

Article Menu

Kolmogorov–Arnold Network for Predicting CO₂ Corrosion and Performance Comparison with Traditional Data-Driven Approaches

Abstract

Share and Cite

Article Metrics

Article Access Statistics

Further Information

Guidelines

MDPI Initiatives

Follow MDPI