Interpretable Pipeline Burst Pressure Prediction Model Based on PCA-HHO-Xgboost

Pipeline transportation plays a crucial role in long-distance oil and gas transport today, which could pose a threat to the lives and properties of local residents if damage occurs. Thus being able to accurately predict the burst pressure of corroded pipelines has a vital role to play in the life-cycle operational management of pipelines. This paper proposes an interpretable model for predicting pipeline burst pressure. Firstly, the Principal Component Analysis (PCA) is used to reduce the dimensions of the original features, obtaining new features that are more informative. Subsequently, Xgboost is employed to predict the pipeline burst pressure, with the Harris Hawks Optimization (HHO) then utilized to optimize the hyperparameters of Xgboost. Finally, the SHapley Additive exPlanations (SHAP) is utilized to interpret the prediction outcomes of PCA-HHO-Xgboost, clarifying the impact of each feature on the pipeline burst pressure.