Geotechnical information forms the cornerstone of geotechnical engineering,playing a vital role in the design,construction,protection,and mitigation of structures by providing critical insights into geomaterials and s...Geotechnical information forms the cornerstone of geotechnical engineering,playing a vital role in the design,construction,protection,and mitigation of structures by providing critical insights into geomaterials and subsurface characteristics.Traditionally,drilling has been the primary method for acquiring subsurface data.Nevertheless,in recent decades,non-intrusive methods such as electrical resistivity surveys,complemented by machine learning(ML)techniques,have emerged as effective tools for comprehensive geotechnical site investigations.This study aims to integrate two-dimensional(2D)electrical resistivity imaging(ERI)with the standard penetration test(SPT)to categorize geomaterials employing k-means clustering analysis(KMCA),facilitating a better understanding of subsurface characteristics.A power equation is developed by correlating soil resistivity with N₆₀-value along a specific 2D line,supported by borehole data.This equation is subsequently applied to derive the estimate of N₆₀-value in areas of the slope where SPT data could not be directly obtained.KMCA proves to be a robust and versatile tool,enabling the classification of geomaterials into three primary clusters corresponding to different competency levels based on SPT measurements.These clusters include low-competency materials(loose sand),moderate-competency materials(medium-stiff clay),and high-competency materials(stiff-hard clay).The application of the power equation successfully estimates the N₆₀-value in areas lacking borehole data,achieving a high level of accuracy with a coefficient of determination(R²)of 0.9467,a mean absolute error(MAE)of 3.94,and a root mean squared error(RMSE)of 5.21.Consequently,this approach demonstrates the potential of integrating geophysical surveys with ML techniques to enhance subsurface characterization and extend the reach of traditional geotechnical methods.展开更多
文摘Geotechnical information forms the cornerstone of geotechnical engineering,playing a vital role in the design,construction,protection,and mitigation of structures by providing critical insights into geomaterials and subsurface characteristics.Traditionally,drilling has been the primary method for acquiring subsurface data.Nevertheless,in recent decades,non-intrusive methods such as electrical resistivity surveys,complemented by machine learning(ML)techniques,have emerged as effective tools for comprehensive geotechnical site investigations.This study aims to integrate two-dimensional(2D)electrical resistivity imaging(ERI)with the standard penetration test(SPT)to categorize geomaterials employing k-means clustering analysis(KMCA),facilitating a better understanding of subsurface characteristics.A power equation is developed by correlating soil resistivity with N₆₀-value along a specific 2D line,supported by borehole data.This equation is subsequently applied to derive the estimate of N₆₀-value in areas of the slope where SPT data could not be directly obtained.KMCA proves to be a robust and versatile tool,enabling the classification of geomaterials into three primary clusters corresponding to different competency levels based on SPT measurements.These clusters include low-competency materials(loose sand),moderate-competency materials(medium-stiff clay),and high-competency materials(stiff-hard clay).The application of the power equation successfully estimates the N₆₀-value in areas lacking borehole data,achieving a high level of accuracy with a coefficient of determination(R²)of 0.9467,a mean absolute error(MAE)of 3.94,and a root mean squared error(RMSE)of 5.21.Consequently,this approach demonstrates the potential of integrating geophysical surveys with ML techniques to enhance subsurface characterization and extend the reach of traditional geotechnical methods.
