Application of Principal Component Analysis in Prediction of Penetration Rate of TBM Using Artificial Neural Networks

Performance analysis and accurate prediction of Penetration Rate (PR) of a TBM have been the ultimate goals of many research works. A reliable prediction of a TBM performance is necessary in budget control and also time schedule planning in underground excavation projects. This research aims to investigate the application of Principal Component Analysis (PCA) in prediction of penetration rate of TBM using Artificial Neural Networks (ANN)، which has not been used and reported in previous studies in this field of study. PCA is a statistical procedure that uses an orthogonal transformation to convert a set of observations of possibly correlated variables into a set of values of linearly uncorrelated variables called principal components. In this study، the obtained data from 10 KM of excavated Zagros tunnel project in Iran are used to predict PR of a TBM using ANN modeling. For predicting PR of TBM، mechanical properties of intact rock and rock masses and also operational parameters such as recorded values of torque and thrust are required. Various simulations with and without PCA have been conducted to achieve the optimum network for predicting PR. The obtained results of simulation show that the network with PCA is more appropriate and the network with 11 principal components as input parameters and 16 neurons in hidden layer is the optimum structure for predicting of PR in this study. The results of developed ANN method show that the optimum network is very efficient for predicting the PR in Zagros tunnel.