Intelligent Prediction of Tensile Strength of Welds in X70 Pipeline Steel Under Small Sample Gas-phase Hydrogen Filling Conditions Using IWOA-GPR

The safe operation of hydrogen-enriched natural gas pipelines (HBNG) requires an accurate understanding of the mechanical performance degradation laws of in-service pipeline materials under hydrogen environments. Currently, the experimental data related to the welding materials of in-service X70 pipelines is still relatively limited. The performance degradation behavior under the combined effects of multiple factors such as hydrogen partial pressure, total pressure, and prehydrogening time still lacks accurate characterization and prediction. Therefore, this study conducted slow strain rate tensile (SSRT) tests on in-service X70 pipeline welding materials and used the nitrogen–hydrogen mixed gas, which has been verified for equivalence, instead of the methane-hydrogen mixed gas for gas-phase hydrogen filling, to obtain tensile strength data under different hydrogen partial pressures, total pressures, and prehydrogening time conditions. On this basis, a Gaussian process regression (GPR) prediction model with hydrogen partial pressure, total pressure, and prehydrogening time as input was constructed. To overcome the problem that the empirical setting of GPR hyperparameters is prone to falling into local optimization under small-sample nonlinear data conditions, the improved whale optimization algorithm (IWOA) was introduced to conduct global optimization of the GPR kernel function hyperparameters. The initial population distribution was optimized through Tent mapping, and the global search and local development balance ability of the algorithm was improved by using a Sigmoid nonlinear convergence factor. The results show that the IWOA-GPR model has an RMSE of 3.40 MPa, MAE of 2.93 MPa, and MAPE of 0.41% on the test set; compared with WOA-GPR, Dung Beetle Optimizer-GPR (DBO-GPR), IWOA-KRR, IWOA-XGBoost, and IWOA-SVR, the maximum reduction in MSE and RMSE is 87.5% and 64.6%, respectively, and the accuracy and stability advantages are significant. Further SHAP sensitivity analysis based on the 24 h hydrogen precharging condition indicates that hydrogen partial pressure is the dominant controlling factor (SHAP = 10.11). This study provides an effective data-driven method for predicting the mechanical property degradation of hydrogen-containing pipeline steel under small sample conditions.

Authors

Institutions

Publication Details

Journal
ACS Omega
Published
2026-11-18
DOI
https://doi.org/10.1021/acsomega.6c08413
Primary Topic
Hydrogen embrittlement and corrosion behaviors in metals
Type
article
Field-Weighted Citation Impact
0.00
Controls
|||
ALL TIME
JAN
FEB
MAR
APR
MAY
JUN
JUL
AUG
SEP
OCT
NOV
article

Intelligent Prediction of Tensile Strength of Welds in X70 Pipeline Steel Under Small Sample Gas-phase Hydrogen Filling Conditions Using IWOA-GPR

Song Deng, Jiuqing Ban, Qiang Liu, Xiaopeng Yan et al.
5 links
ACS Omega
Hydrogen embrittlement and corrosion behaviors in metals
article

Intelligent Prediction of Tensile Strength of Welds in X70 Pipeline Steel Under Small Sample Gas-phase Hydrogen Filling Conditions Using IWOA-GPR

Song Deng, Jiuqing Ban, Qiang Liu, Xiaopeng Yan, Weilin Guo, Xin Wang, Wei Yang
article en

Abstract

The safe operation of hydrogen-enriched natural gas pipelines (HBNG) requires an accurate understanding of the mechanical performance degradation laws of in-service pipeline materials under hydrogen environments. Currently, the experimental data related to the welding materials of in-service X70 pipelines is still relatively limited. The performance degradation behavior under the combined effects of multiple factors such as hydrogen partial pressure, total pressure, and prehydrogening time still lacks accurate characterization and prediction. Therefore, this study conducted slow strain rate tensile (SSRT) tests on in-service X70 pipeline welding materials and used the nitrogen–hydrogen mixed gas, which has been verified for equivalence, instead of the methane-hydrogen mixed gas for gas-phase hydrogen filling, to obtain tensile strength data under different hydrogen partial pressures, total pressures, and prehydrogening time conditions. On this basis, a Gaussian process regression (GPR) prediction model with hydrogen partial pressure, total pressure, and prehydrogening time as input was constructed. To overcome the problem that the empirical setting of GPR hyperparameters is prone to falling into local optimization under small-sample nonlinear data conditions, the improved whale optimization algorithm (IWOA) was introduced to conduct global optimization of the GPR kernel function hyperparameters. The initial population distribution was optimized through Tent mapping, and the global search and local development balance ability of the algorithm was improved by using a Sigmoid nonlinear convergence factor. The results show that the IWOA-GPR model has an RMSE of 3.40 MPa, MAE of 2.93 MPa, and MAPE of 0.41% on the test set; compared with WOA-GPR, Dung Beetle Optimizer-GPR (DBO-GPR), IWOA-KRR, IWOA-XGBoost, and IWOA-SVR, the maximum reduction in MSE and RMSE is 87.5% and 64.6%, respectively, and the accuracy and stability advantages are significant. Further SHAP sensitivity analysis based on the 24 h hydrogen precharging condition indicates that hydrogen partial pressure is the dominant controlling factor (SHAP = 10.11). This study provides an effective data-driven method for predicting the mechanical property degradation of hydrogen-containing pipeline steel under small sample conditions.

ACS Omega
Southwest Petroleum University (CN), Changzhou University (CN), PetroChina Southwest Oil and Gas Field Company (China) (CN)
Industry, innovation and infrastructure
Hydrogen embrittlement and corrosion behaviors in metals
AI Navigator

Ask Laika to Summarize, Analyze, and Connect papers live on the map.

Summarize Papers & Methodologies

Extract key findings, datasets, and comparative methods across publications.

Benchmark Rankings & Visual Analytics

Rank top research institutions, authors, funders, topics, and journals by Field-Weighted Citation Impact (FWCI) and paper volume with instant charts.

Connect Distant Disciplines

Bridge topological clusters on the map to find hidden collaborative intersections.

Laika access coming soon