A physics-informed machine learning framework for predicting the failure pressure of buried PVC pipelines with axial cracks

Due to the coupling effect of internal pressure, external soil load and crack geometry, accurately predicting the failure pressure of cracked buried pipelines remain a critical challenge. This study develops a physics-informed neural network. By embedding physical constraints into the loss function, mechanical principles are integrated into the training process of the model. Furthermore, the Bayesian optimisation strategy is adopted to obtain the optimal hyperparameters of the model automatically. The proposed model integrates geometric, material and load parameters and is capable of accurately predicting the failure pressure of cracked buried pipelines under complex underground conditions. Through performance testing and comparison with traditional data-driven models, this model has achieved more satisfactory prediction accuracy and robustness. Meanwhile, this model exhibits excellent generalization capability on small sample datasets. The results show that combining physics-informed with machine learning can provide a reliable and effective method for the health assessment of pipeline structures with cracks.