Assessing the strength of deep-sea surface ultrasoft sediments with T-bar penetration: A machine learning approach

Abstract

With the expansion of marine resource utilization, geological disaster prediction, and ecological protection in deep-sea regions, accurate identification of deep-sea engineering geological environments has become paramount, highlighting the significance of understanding the mechanical characteristics of deep-sea surface ultrasoft sediments (DSUSs). However, DSUSs typically exhibit low strength, high compressibility, and significant fluidity, making sampling for indoor strength testing a challenging task. On-site testing employing a T-bar penetrometer has become increasingly prominent for assessing DSUS strength, but the complexity of influencing parameters limits the application of this valuable method. This study utilizes a validated computational fluid dynamics (CFD) approach to model the complete process of T-bar penetration into a DSUS. The aim is to understand the quantitative effects and corresponding mechanisms of various significant complex parameters, including the undrained shear strength, dimensionless penetration depth, rate effect, and interface contact condition, on the dimensionless penetration resistance coefficient. Additionally, a machine learning model based on the random forest algorithm is introduced to establish a multiparameter evaluation method for the dimensionless penetration resistance coefficient. The findings illuminate the intricate interactions and effective evaluation of these factors when evaluating DSUS strength using a T-bar penetrometer, addressing the above challenges in marine engineering geology and the environment.