Development of a comprehensive model to evaluate the foam consumption and the advance rate in EPB tunnel boring machine

In this study, in order to investigate and evaluate the performance of the EPB TBM, the parameters of progress rate and foam consumption for ground improvement have been used. To evaluate the performance of EPB tunnel boring machine in this research, drilling data were used from 4 databases namely Qom metro line one project, Shiraz metro line two project (east and west tunnel), Isfahan metro line two project and Isfahan metro line one project which have been EPB TBM excavated. Based on this database, models based on statistical techniques including regression analysis (simple regression, linear multivariate (MLR) and nonlinear (MNLR) regressions) and support vector machine least squares (LS-SVM) algorithm were developed to evaluate the foam consumption and the advance rate. Then, in order to study more accurately and better the experimental and intelligent models presented to predict the foam consumption and the advance rate, statistical indicators such as the coefficient of determination (R2), the statistical values ​​of the root mean square of the normalized error (NRMSE) and error variance (VAF) for each developed model were calculated using the test data set. Comparison of the calculated values of R2, VAF and NRMSE for the models developed to predict the foam consumption and the advance rate showed that the proposed model based on the LS-SVM algorithm has remarkable accuracy compared to other models, in predicting the amount of foam consumption and the advance rate of the EPB tunnel boring machine. The values of R2, VAF and NRMSE indices for the developed model in the prediction of the amount of foam consumption are 0.945, 94.356 and 0.237 respectively, and in the prediction of the advance rate are equal to 0.741, 74.071 and 0.149, respectively.  Due to the high accuracy of predicting the amount of foam consumption (Qt) and the advance rate (AR), the models which are developed on the LS-SVM algorithm basis, can play a significant role as applied models in the mechanized drilling industry.