Pipejacking clogging detection in soft alluvial deposits using machine learning algorithms

• Potential for AI techniques to detect clayey clogging in pipejacking is explored. Data decomposition into feature-based sub-series accentuates their features. The use of slurry density, torque and speed can be useful detecting clogging. ‘Clogging’ a common issue encountered during tunnelling soils which impede tunnel excavation, cause unplanned downtimes lead significant additional project costs. Clogging result drastic reduction performance due reduced jacking speeds the time needed cleaning if it cannot fully mitigated. data acquired by modern boring machines (TBMs) have grown significantly recent years presenting substantial opportunity application data-driven artificial intelligence (AI) techniques. In this study, baseline assessment slurry-supported performed using combination TBM parameters semi-empirical diagrams proposed literature. potential one-class support vector (OCSVM), isolation forest (IForest) robust covariance (Robcov) assess tendency then explored work. approach applied case history Taipei, Taiwan, involving soft alluvial deposits. results highlight an exciting pipejacking.