Risk Assessment in Energy Infrastructure Installations by Horizontal Directional Drilling Using Machine Learning

Nowadays we can observe a growing demand for installations of new gas pipelines in Europe. A large number them are installed using trenchless Horizontal Directional Drilling (HDD) technology. The aim this work was to develop and compare machine learning models dedicated risk assessment HDD projects. data from 133 projects eight countries the world were gathered, profiled, preprocessed. Three models, logistic regression, random forests, Artificial Neural Network (ANN), developed predict overall project outcome (failure free installation or likely fail), occurrence identified unwanted events. best performance terms recall accuracy achieved ANN model, which proved be efficient, fast robust predicting risks Machine applications proposed enabled eliminating involvement group experts process therefore significantly lower costs associated with process. Future research may oriented towards developing comprehensive management system, will enable dynamic taking into account various combinations mitigation actions.