Cross-project prediction for rock mass using shuffled TBM big dataset and knowledge-based machine learning methods

Extensive research has confirmed the successful prediction of rock mass quality in tunnel boring machine (TBM) construction using learning methods based on big data collected during process. However, developed model cannot be applied to a new project owing different mechanical and environmental features involved projects. This study tries combine datasets three TBM projects whose cutterhead diameters are 5.2, 7.9 9.8 m, respectively. In this study, focused predictions binary system was implemented unified dataset by adding diameter disc cutter number as attributes into input. The process consists of: (1) individual for respective datasets, (2) shuffled containing randomly distributed information from projects, (3) crossed aimed at validating that algorithm can produce with equally acceptable accuracies those obtained learning. It is anticipated more joining cross-project learning, we will able develop suitable wide range numbers beginning excavation.