Ehsan Pirhadi - Department of Mining, Science and Research Branch, Islamic Azad University, Tehran,Iran
Kourosh Shahriar - Department of Mining and Metallurgical Engineering, Amirkabir University of Technology, Tehran, Iran

Rate of penetration (ROP) of a tunnel boring machine (TBM) in a rock environment is generally akey parameter for the successful accomplishment of a tunneling project. To develop the proposedmodels, the database that is composed of intact rock properties including uniaxial compressivestrength (UCS), Brazilian tensile strength (BTS), and peak slope index (PSI), and also rock massproperties including distance between planes of weakness (DPW) and the alpha angle (α) are inputas dependent variables and the measured ROP is chosen as an independent variable. In this study, theTehran-Karaj water conveyance tunnel located in the province of Alborz has been chosen to beinvestigated. Initially data were collected and then effective parameters on the penetration rate weredetermined. Support vector machine (SVM) is a novel machine learning technique usually consideredas a robust artificial intelligence method in classification and regression tasks. To investigate thesuitability of this approach, the predictions by SVM have been compared with multi variableregression (MVR), too. The accuracy of the prediction models is measured by the coefficient ofdetermination correlation coefficient (R2) between predicted and observed yield employing 5-foldcross-validation schemes. Model statistical parameters show that there is a very good relation betweenROP and the model variables with a R2=0.75 for MVR and 0.99 for SVM. Also, the squaredcorrelation coefficient for prediction set was achieved 0.65 for MVR and 0.98 for SVM.

rate of penetration; tunnel boring machine; support vector machine