Rock mass classification method applying neural networks to minimize geomechanical characterization in underground Peruvian mines

• Autores: Brousset, Julyans, Pehovaz, Humberto, Quispe, Grimaldo, Raymundo, Carlos, Moguerza, Javier M.
• Formato: artículo
• Fecha de Publicación: 2023
• Institución: Universidad Peruana de Ciencias Aplicadas
• Repositorio: UPC-Institucional
• Lenguaje: inglés
• OAI Identifier: oai:repositorioacademico.upc.edu.pe:10757/668730
• Enlace del recurso: http://hdl.handle.net/10757/668730
• Nivel de acceso: acceso abierto
• Materia: Artificial intelligence; Artificial neural networks; Geomechanical classification; Geomechanical model; Uncertainty; Classification; Rock mass; Underground mining; Geological alterations; Input/output data; Training; Real-time; RMR index; Precise classification results

This research aims to enhance the classification of the rock mass in underground mining, a common problem due to geological alterations that do not fit existing methods. Artificial neural networks are proposed as a solution, which use input/output data to learn and solve problems. The process involves gathering data on rock properties and training the neural networks to identify and classify various types of rock. Once trained, the neural networks can classify the rock mass in real-time during mine design and progression, adapting to different rock types with a low margin of error of 0.279% in determining the RMR index. This research overcomes the limitations of current classification methods, providing a more accurate and reliable solution for the classification of the rock mass in underground mining. In summary, artificial neural networks are utilized to improve the classification of rock mass in underground mining by adapting to geological changes and providing precise classification results.