Prediction of Soil Shear Strength Parameters Using Combined Data and Different Machine Learning Models

Soil shear strength is an important indicator of soil erosion sensitivity and the tillage performance cultivated layer. Measuring at a field scale difficult, time-consuming, costly. This study proposes new method to predict parameters (cohesion internal friction angle) by combining cone penetration test (CPT) data properties. A portable CPT measuring device with two pressure sensors was designed collect in farmland, namely tip resistance, side pressure. Direct tests were performed laboratory determine for 83 collection points. Two easily available properties (water content bulk density) determined via oven-drying method. Using as predictors, three machine learning (ML) models built predicting cohesion angle, including backpropagation neural network (BPNN), partial least squares regression (PLSR), support vector (SVR). The prediction each model evaluated using coefficient determination (R2), root-mean-square error (RMSE), relative (RE). results suggested that among all models, BPNN most suitable cohesion, SVR best angle. Thus, our findings provide foundation convenient low-cost measurement parameters.