Single Valued Neutrosophic Number Ensemble Learning Model for Stability Classification of Open Pit Mine Slopes

Abstract

China's open pit mining industry faces the dual challenge of increasing production and 
preventing disasters. In order to ensure the safe exploitation of mineral resources, the stability of 
slopes must be assessed. In light of the fact that over 1,950 landslide accidents have occurred over 
the past decade, accounting for 15% of all safety incidents, the evaluation of slope stability has 
become a critical research focus in the fields of geo-resources and geo-engineering. Traditional 
slope stability evaluation methods rely on empirical tools and the expertise of professionals to 
assess slope stability. In contrast, machine learning (ML) methods offer a more comprehensive 
approach, analyzing the intricate features present in diverse sampling data. As a novel extension of 
ML, this paper presents a single-valued neutrosophic number-based ensemble learning (SVNN-EL) 
model. This model employs binary coding groups (1, 0, 0), (0, 1, 0) and (0, 0, 1) to express the 
learning outcomes of the slope stability, quasi-stability and instability statuses. Subsequently, a 
similarity measure is employed to determine the classification results of slopes. Finally, the 
proposed SVNN-EL model is applied to a case study in Yunnan province, China, the proposed 
model's four performance metrics, namely accuracy, precision, recall, and F1-score (the harmonic 
mean of precision and recall), are 0.915, 0.894, 0.948, and 0.921, respectively. A comparison with the 
k-nearest neighbor, support vector machine and random forest methods reveals that the 
performance metrics of the proposed SVNN-EL model are superior to those of existing methods. 

DOI: 10.5281/zenodo.13931996