Resistivity of Frozen Silty Clay Using an RC Circuit-Based Testing System and MARS Prediction Model

The significant contrast in electrical resistivity between frozen and unfrozen soils makes resistivity tomography a widely used technique in surveys of frozen soil engineering. To investigate the electrical resistivity characteristics of frozen soils, this paper proposes an indoor frozen soil resistivity testing system based on an RC circuit. The system evaluates soil resistivity by monitoring voltage changes during the charge-discharge cycles of a capacitor in an RC circuit under alternating current. The study focuses on the silty clay in Harbin and the Xiaoxingan Mountain regions. Subsequently, the proposed resistivity testing system is used to evaluate soil resistivity under various porosity, temperature, and unfrozen water content conditions. The results indicate that the resistivity of frozen soil rises with increasing porosity due to the increment in pore ice content. Moreover, the performance of existing frozen soil electrical resistivity models is assessed using experimental data. Further, using experimental data, multivariate adaptive regression splines (MARS), a machine learning method, is employed to establish a resistivity prediction model for frozen soil. A comparative analysis is conducted between the predictive formula derived from the MARS model, and its performance is compared with existing models. This study contributes to frozen silty clay electrical resistivity testing and understanding the electrical properties of frozen silty clay, which is crucial for applications such as permafrost site exploration.