A slope stability classification model based on the matrix energy of neutrosophic confidence cubic sets for multi-point sampling data of slopes

Abstract

Open-pit mining, the predominant form of mineral resource extraction in China, faces the simultaneous challenges of increasing production demands and hazard mitigation. Consequently, the assessment of slope stability in open-pit mines is imperative and vital to ensure the integrity and safety of the nation’s mineral resources. Conventional slope stability evaluation models and methods encounter a significant challenge in adequately expressing sampling information due to uncertainty, incompleteness, and imprecision in the sampling process. Therefore, we need to develop a novel approach to convey the sampling information more effectively. Based on the integration of Neutrosophic Confidence Cubic Sets (NCCSs) and matrix energy, this study first proposes the NCCS matrix and matrix energy based on a confidence level in a multi-valued matrix scenario and then constructs a classification model using the NCCS Matrix Energy (NCCSME) and its score function. Next, a case study, involving 212 actual slope samples in the Lanping open-pit mine, located in Yunnan, China, is provided to demonstrate the validity and reasonability of the proposed classification model. Additionally, the classification results are compared with seven common Machine Learning (ML) methods, including K-Nearest Neighbors (KNN), Support Vector Machine (SVM), Multi-Layer Perceptron (MLP), Random Forest (RF), Linear Regression (LR), Extreme Gradient Boosting (XGBoost), and Native Bayes Classifier (NBC). The performance metrics of the proposed model are presented as follows: 0.953 for accuracy (Acc), 0.950 for precision, 0.955 for recall, and 0.952 for F1-score. These metrics are better than those obtained by the common ML methods. In summary, the main contribution of this study is that the proposed classification model using NCCSME can address both the challenges of expressing and processing multi-point sampling data for each slope and the classification of slope stability into stable, failed/unstable, and quasi-stable statuses based on a confidence level