| Title |
Symbolic Regression-Based Prediction of Shear Strength Parameters for Rocks |
| Authors |
양예림(Yang, Yerim);최항석(Choi, Hangseok);권기범(Kwon, Kibeom) |
| DOI |
https://doi.org/10.12652/Ksce.2026.46.3.0237 |
| Keywords |
전단강도정수; 점착력; 내부마찰각; 암석; 기호회귀 Shear strength parameter; Cohesion; Angle of internal friction; Rock; Symbolic regression |
| Abstract |
Accurate estimation of rock shear strength parameters is critical for ensuring the long-term stability of geotechnical infrastructures. However, conventional approaches are often constrained by prior assumptions regarding the functional form or by limited interpretability due to the black-box nature of the models. In this study, symbolic regression was employed to derive explicit predictive equations for the shear strength parameters: cohesion (c) and the angle of internal friction (φ). A comprehensive database comprising 199 rock samples was compiled, including P-wave velocity (Vp), density (ρ), uniaxial compressive strength (UCS), and tensile strength(TS) along with the corresponding target parameters (c and φ). Correlation analysis revealed that c is strongly and positively correlated with Vp, UCS, and TS, while φ exhibits the strongest positive correlation with ρ, showing moderate correlations with the other variables. Based on these findings, univariate predictive equations for c were derived using Vp, UCS, and TS as individual input variables, whereas a multivariate predictive equation for φ was developed using Vp and ρ. The proposed equations demonstrated superior predictive performance, achieving R2 values of 0.93-0.94 for c and 0.83 for φ. Furthermore, comparative evaluations against eleven established empirical models confirmed that the proposed equations provide enhanced performance, validating their practical applicability in geotechnical design. These results indicate that symbolic regression is an effective approach for deriving interpretable and accurate predictive equations for rock shear strength parameters. |