Answers are grounded in this paper and cite their sources.
This study harnesses Artificial Neural Networks to predict saturated unit weight values utilizing Cone Penetration Tests parameters. Current CPT-based models overestimate unit weight values for organic soils. Although methods to predict unit weight using machine learning algorithms such as ANNs exist, this study aims to improve the prediction accuracy of organic soils by incorporating additional parameters. The ANN model developed herein consistently achieves low normalized mean absolute errors (< 7%) and high coefficients of determination (R2 ~ 0.70) across diverse soil types, including organics and inorganics. The practical implications of more accurate γsat predictions in slope stability analyses, particularly for levee embankments exemplified by the West Shore Lake Pontchartrain project, are also assessed. The study emphasizes the pivotal role of accurate γsat values in these analyses, highlighting the ANN model's advantage of conducting stability assessments along the entire levee alignment from CPTu data, in contrast to traditional practices confined only to soil boring locations. Through database expansion, a comprehensive tool applicable to projects across the United States and beyond can be developed. This research underscores the transformative potential of advanced machine learning techniques for enhancing traditional geotechnical practices by leveraging available data to determine design parameters in geotechnical engineering.
No takes yet. Share an insight, caveat, or question.
Ulloa et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: