A Machine Learning-Integrated Model for the Installation of Offshore Cables in Sandy Soils with Ploughing

This study stems from the increasing demand for reliable and efficient submarine cable installations, crucial for global communication and energy infras-tructure. Cables are usually buried in the seabed, with vessel-driven mechanical trenchers often preferred over jetting systems. Accurate prediction of the trench-ing force becomes essential for optimizing the installation process. Several models have been proposed in the literature expressing the force as a function of soil prop-erties, plough geometry, trenching depth and velocity. However, these formulations are often oversimplified and inaccurate. Their applicability is further restricted by the quantity and quality of geotechnical data available in offshore environments. This last issue was recently addressed by Robinson et al. (2021). In their work they proposed a model whose parameters can be readily determined from in-situ Cone Penetration Tests (CPTs), avoiding the use of field-derived empirical coefficients. Building upon this framework, a series of design charts is proposed here relating the speed effect to CPT measurements and soil properties that are unaffected by sampling-induced disturbance. These charts have been derived from a supervised regression-based machine learning (ML) algorithm.