Predictive modelling

Random forest

Engineering

Statistics

Support vector machine

Data mining

Computer science

Parametric model

Mathematics

Pipeline (software)

Artificial intelligence

Parametric statistics

Programming language

Artificial neural network

Machine learning

International Journal of Pressure Vessels and Piping

Establishing an accurate model to predict burst pressure is desired, which has been developed for decades. Although various models have been developed, errors unavoidably appear in the prediction of burst pressures because of the uncertainty in both input variables and nonlinear relationship of such variables to the burst pressure. Consequently, machine learning models, which is a data-driven approach, are potential alternatives. In this paper, various machine learning models such as Random Forest, Support Vector Machine, and Artificial Neural Network are examined to predict the burst pressure, gathering databases available in the literature. The applications of these models are investigated to identify the advantages and limitations of the models. The machine learning models showed a significant improvement in the prediction of the burst pressures compared to the available reference models. However, some drawbacks of the models should be carefully considered, including an increase of error with the unfamiliar data and the fluctuations within the overall trend in the parametric study.

Predicting pipeline burst pressures with machine learning models