PREDICTION METHOD OF SOIL CONSTANT BY USING CONE PENETRATION TEST AND ARTIFICIAL INTELLIGENCE TECHNOLOGY

Abstract

 Soil investigation to obtain the N value and soil constants of the seafloor ground require a workbench, such as the Self-Elevating Platform (SEP) ship, as well are strongly affected by meteorological and oceanographic conditions. Therefore, a simple investigation method such as a cone penetration test is required. In the cone penetration test, it is possible to predict the soil constants using empirical conversion formulas. However, the conversion formulas need minor adjustments for different soil properties and investigation locations. In this study, we will improve the prediction accuracy of the soil constants (N value, F_c) by utilizing the artificial intelligence technology (AI). The results of the study suggested that the use of AI could improve the estimation accuracy. When applying the AI technology, a number of existing data is required for pretraining the prediction method.