Multi-Parameter Maximum Corrosion Depth Prediction Model for Buried Pipelines Based on GSCV-XGBoost
Niannian Wang, Liuyang Song, Hongyuan Fang, Bin Li, Fuming Wang
Abstract
Corrosion is one of the most common types of damage in buried oil and gas pipelines. Corrosion leaks can cause serious accidents and can be harmful to pipelines during service. The maximum corrosion depth of an oil and gas pipeline is an important indicator for assessing the remaining strength of the pipeline. An accurate prediction of the maximum corrosion depth is important for the safe operation of pipelines. Machine learning has been shown to perform well in predictive assessment efforts. However, previous studies have rarely considered the effects of corrosion characterization and parameter optimization simultaneously. In this study, a multi-parameter maximum corrosion depth prediction model for pipelines based on GSCV-XGBoost is proposed, which can be applied to real projects. The model performs feature extraction on the pipeline dataset through Pearson correlation analysis, identifies the parameters that contribute more to the maximum corrosion depth, and predicts the maximum corrosion depth of the pipeline using an optimized machine learning model. The machine learning model used in this study was obtained by optimizing the XGBoost model using the GirdSearchCV method. That is, the optimal hyperparameter combination of the model was obtained by 10-fold cross-validation and grid searching. The prediction results were compared with those of five common machine learning models. The conclusions show that the GSCV-XGBoost model performs the best in predicting the maximum corrosion depth of the pipeline with the smallest error. The R <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sup> and Root Mean Square Error(RMSE) scores for the test set were 0.9886 and 0.2057, respectively. The prediction accuracy was improved by 34.59% over that of the conventional XGBoost model.