An application of adaptive neuro fuzzy inference system for estimating the uniaxial compressive strength of certain granitic rocks from their mineral contents


Yesiloglu-Gultekin N., Sezer E. A., Gokceoglu C., Bayhan H.

EXPERT SYSTEMS WITH APPLICATIONS, vol.40, no.3, pp.921-928, 2013 (SCI-Expanded) identifier identifier

  • Publication Type: Article / Article
  • Volume: 40 Issue: 3
  • Publication Date: 2013
  • Doi Number: 10.1016/j.eswa.2012.05.048
  • Journal Name: EXPERT SYSTEMS WITH APPLICATIONS
  • Journal Indexes: Science Citation Index Expanded (SCI-EXPANDED), Scopus
  • Page Numbers: pp.921-928
  • Keywords: Granitic rock, Nonlinear multiple regression, Adaptive neuro fuzzy inference system, Uniaxial compressive strength, PETROGRAPHIC IMAGES, EDGE-DETECTION, PREDICTION, MODULUS, ANFIS, ELASTICITY, REGRESSION, SANDSTONES, NETWORKS, MODELS
  • Hacettepe University Affiliated: Yes

Abstract

The uniaxial compressive strength (UCS) of rocks is an important intact rock parameter, and it is commonly used for various engineering applications. This parameter is mainly controlled by the mineralogical and textural characteristics of rocks. In this study, a soft computing method, an adaptive neuro-fuzzy inference system (ANFIS), was employed to estimate UCS from the mineral contents of certain granitic rocks selected from Turkey; nonlinear multiple regression analysis was then employed to validate these estimations. Five nonlinear multiple regressions and ANFIS models were constructed with three inputs: quartz, orthoclase and plagioclase. To determine the optimal model, various performance indices (R, values account for and root mean square error) were determined, and the model obtained from dataset #3 was selected as the optimal model. The coefficients of correlation for the nonlinear multiple regression and ANFIS models were 0.87 and 0.91, respectively. Thus, both models yielded acceptable results, and the ANFIS is a suitable method for estimating the UCS of rocks. (C) 2012 Elsevier Ltd. All rights reserved.