A novel systematic and evolved approach based on XGBoost-firefly algorithm to predict Young’s modulus and unconfined compressive strength of rock

• Published in: Engineering with computers Vol. 38; no. Suppl 5; pp. 3829 - 3845
• Main Authors: Cao, Jing, Gao, Juncheng, Nikafshan Rad, Hima, Mohammed, Ahmed Salih, Hasanipanah, Mahdi, Zhou, Jian
• Format: Journal Article
• Language: English
• Published: London Springer London 01.12.2022; Springer Nature B.V
• Subjects: Algorithms; CAE) and Design; Calculus of Variations and Optimal Control; Optimization; Civil engineering; Classical Mechanics; Compressive strength; Computer Science; Computer-Aided Engineering (CAD; Control; Feasibility; Grain size; Heuristic methods; Laboratories; Machine learning; Math. Applications in Chemistry; Mathematical and Computational Engineering; Modulus of elasticity; Neural networks; Original Article; R&D; Radial basis function; Regression analysis; Research & development; Rock properties; Root-mean-square errors; Stone; Support vector machines; Systems Theory; Tensile strength; Tunnels; XGBoost; Firefly algorithm; Machine learning; Rock properties
• ISSN: 0177-0667; 1435-5663
• DOI: 10.1007/s00366-020-01241-2

To design the tunnel excavations, the most important parameters are the engineering properties of rock, e.g., Young’s modulus (E) and unconfined compressive strength (UCS). Numerous researchers have attempted to propose methods to estimate E and UCS indirectly. This task is complex due to the difficulty of preparing and carrying out such experiments in a laboratory. The main aim of the present study is to propose a new and efficient machine learning model to predict E and UCS. The proposed model combines the extreme gradient boosting machine (XGBoost) with the firefly algorithm (FA), called the XGBoost-FA model. To verify the feasibility of the XGBoost-FA model, a support vector machine (SVM), classical XGBoost, and radial basis function neural network (RBFN) were also employed. Forty-five granite sample sets, collected from the Pahang-Selangor tunnel, Malaysia, were used in the modeling. Several statistical functions, such as coefficient of determination ( R 2 ), mean absolute percentage error (MAPE) and root mean square error (RMSE) were calculated to check the acceptability of the methods mentioned above. A review of the results of the proposed models revealed that the XGBoost-FA was more feasible than the others in predicting both E and UCS and could generalize.