An integrated framework combining single-valued neutrosophic feature processing and machine learning for predicting the uniaxial compressive strength of rocks
Year: 2026
Type: article
Abstract: The uniaxial compressive strength (UCS) of rocks is a significant aspect of many rock mechanics investigations as well as a key parameter that is utilized in theoretical and engineering applications. This study introduces a single-valued neutrosophic number (SVNN) based extreme gradient boosting (XGBoost) and ridge regression (SVNN-XGBoost-RR) model in the uncertainty and incompleteness of the input features to enhance the predictability of UCS of rocks. In the application of the given model for the UCS prediction, the 1,190 rock samples used for machine learning, which include three features (P-wave velocity, Point load strength, Schmidt hammer rebound value), are first converted into SVNNs using a neutrosophic method based on true, false, and indeterminate Gaussian membership functions. Subsequently, the proposed model is then trained on 1,080 rock samples and tested on 110 rock samples, including SVNNs of the input features (P-wave velocity, Point load strength, Schmidt hammer rebound value), as well as their UCS output values. The UCS prediction results indicate that the proposed model outperforms the other five machine learning models for both sedimentary and igneous rocks, achieving a coefficient of determination (R 2 ) of 0.9377, a mean absolute error (MAE) of 3.7836 MPa, and a root mean square error (RMSE) of 6.2379 MPa for sedimentary rocks, and an R 2 of 0.8325, MAE of 14.1880 MPa, and RMSE of 17.4990 MPa for igneous rocks. The proposed SVNN-XGBoost-RR model offers effective support for the estimation of rock mass mechanical parameters in the stability analysis and design of geotechnical engineering.
Source: PeerJ Computer Science
Institution: Ningbo University
Language: English
Topic: Rock Mechanics and Modeling
Subfield: Mechanics of Materials
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $1,395
OpenAlex ID: W7235600568
