Predictive Models for Permeability of Cracked Rock Masses Based on Support Vector Machine Techniques
- PDF / 472,700 Bytes
- 9 Pages / 547.087 x 737.008 pts Page_size
- 101 Downloads / 203 Views
(0123456789().,-volV) ( 01234567 89().,-volV)
ORIGINAL PAPER
Predictive Models for Permeability of Cracked Rock Masses Based on Support Vector Machine Techniques Guotao Ma . Zhiming Chao . Kun He
Received: 23 January 2020 / Accepted: 28 August 2020 Ó Springer Nature Switzerland AG 2020
Abstract In this study, a database developed from existing literature about permeability of cracked rock was established. The performance of Support Vector Machine (SVM) combined with optimisation algorithms: Genetic Algorithm (GA) and Particle Swarm Optimisation Algorithm (PSO) in predicting the permeability of cracked rock masses (CRM) is evaluated. Also, the sensitivity analysis of the influence factors to the permeability of CRM is conducted. The results indicate that the hybrid GA–SVM and hybrid PSO–SVM models can accurately predict the permeability of CRM in terms of the statistical performance criteria: Coefficient of Determination R2, Regression Coefficient R and Mean Residual Error (MSE); Additionally, optimisation algorithms: PSO and GA can improve significantly the predictive performance of the SVM model. Based on the sensitivity analysis, crack angle is the most important factor to change the permeability of CRM, followed by confining pressure. Keywords Machine learning Permeability Cracked rock masses Confining pressure SVM
G. Ma School of Engineering, The University of Warwick, Coventry CV4 7AL, UK Z. Chao (&) K. He Research and Development Department, DP Consultation Company, Chengdu 610000, China e-mail: [email protected]
1 Introduction Cracked rock masses (CRM) are frequently encountered in civil and hydraulic engineering projects. Most of existing research efforts about the geotechnical characteristics of CRM had concentrated on their mechanical properties (Chen et al. 2014; Jin et al., 2016; Xiao et al., 2017), while the investigation about the permeability characteristics of CRM is rare. However, due to the presence of crack structure, it provides the major seepage channels for fluid flow while the permeability of rock matrix is extremely low (Fan et al. 2017). Under certain extreme conditions, such as flooding, rainstorm, etc., it is effortless for fluid to penetrate into CRM through the cracks, which decreases significantly the mechnical strength of CRM. In recent years, machine learning techniques have been widely applied in geotechnical engineering and more specific, in rock mechanics (Ceryan 2014; Chou et al. 2016; Gui et al. 2019). SVM, as a powerful tool of machine learning techniques, is often utilised as an effective methodology to predict the properties of rock masses based on lots of influence factors with complex interaction relationship (Gholami et al. 2012; Saeidi et al. 2014). For example, Pham et al. (2018) investigated and compared the performance of four machine learning models, Particle Swarm Optimization (PSO)-Adaptive Network based Fuzzy Inference System, Genetic Algorithm (GA)-Adaptive Network based Fuzzy Inference System (GANFIS), Support
123
Geotech Geol Eng
Vector Regres
Data Loading...
