Application of adaptive neuro-fuzzy inference system and differential evolutionary optimization for predicting rock displacement in tunnels and underground spaces. (February 2023)

• Record Type: Journal Article
• Authors: Zhou, Xiaoguang; Nguyen, Hoang; Trong Hung, Vo; Lee, Chang-Woo; Nguyen, Van-Duc
• Is Part Of: Structures. Volume 48(2023)
• Page Start: 1891
• Page End: 1906
• Publication Date: 2023-02
• Subjects: Stability of tunnels -- Rock mechanics -- Safety analysis -- Intelligence fuzzy system -- Reliability system; Structural engineering -- Periodicals; 624.1
• DOI: 10.1016/j.istruc.2023.01.059
• Languages: English
• ISSNs: 2352-0124

Abstract

Abstract: In this paper, the rock displacement phenomenon was investigated to evaluate and prevent the failure and collapse of tunnels and underground spaces. The historical datasets were then analyzed and used to develop a novel, simpler intelligent model, that provides better accuracy for predicting rock displacement in tunnels/underground spaces, namely the RF-DE-ANFIS model. Whereas the random forest (RF) algorithm played as a robust tool to reduce the number of input variables to make the model's structure is simpler, the differential evolutionary algorithm (DE) optimized the parameters of the adaptive neuro-fuzzy inference system (ANFIS) under 10-folds cross-validation technique for predicting rock displacement based on the selected input variables (by the RF). The models, such as standalone ANFIS, RF-ANFIS (without the optimization of DE), DE-ANFIS (without the reduction of RF), support vector machine (SVM), RF-SVM, and gene express programming (GEP), were also considered and compared with the RF-DE-ANFIS model to demonstrate the simplicity, as well as high efficiency, of the RF-DE-ANFIS model in predicting rock displacement. The obtained results indicated that the RF-DE-ANFIS model provided the most dominant accuracy (∼86 %) with a simpler structure, and the accuracy was improved by 8 % —12 % compared to the other models. Based on the proposed RF-DE-ANFIS model, the rock displacement phenomenon in tunnels/underground spaces can be early warned and proper solutions can be applied to prevent the failure of tunnels/underground spaces.