Principal Component Analysis of Geological and Tunnel Boring Machine Parameters in Hard Rock (Thailand)

Abstract

Tunnel boring machine (TBM) performance prediction is essential for feasibility evaluation and planning purposes. Prediction models involve multiple geological and machine factors from field and laboratory studies. This paper presented an alternative method of feature selection for a less complex yet predictive model. The principal component analysis (PCA) was applied to a multivariate dataset of geological and tunnel boring machine parameters including thrust force, torque, the amount of small-, medium- and large-sized rock chips, quartz content, TBM penetration distance per revolution, and disc cutter wear of Mae Tang to Mae Ngat diversion tunnel project, Chiang Mai, Thailand. Variables with high loading factors from PC1 and PC2 are thrust force, torque, the amount of small- and medium-sized rock chips, and penetration rate. These five features have high contributions to the trends in data variation and should be considered as the most influential parameters when developing TBM performance models.