The present study aims to apply intelligent method to predict settlement of tunnels. This method can be used for settlement prediction of tunnel future levels. Many parameters affect settlement, but some of them are applicable to empirical equations. Prediction of this phenomenon depends on many parameters and the effect of each parameter on the other ones has made the use of empirical methods impossible. In this regard, the capability of Gene Expression Programming (GEP) method are studied for settlement (ST) prediction. The intelligent methods have been studied on the basis of data obtained from 50 tunnels all over the world, which have been excavated using the NATM method with similar soil properties. These parameters were collected from previous research data, which the values of settlement were obtained from numerical modeling (FLAC2Dsoftware). 30 datasets were applied for modeling and 20 datasets were used for appraising its performance. The values of STare predicted by using soil strength parameters (E, C and ø), depth (Z) and diameter (D) of the tunnel. Three equations were also rendered by GEP method. Finally, the results of this method were compared. According to the results, intelligent method are recommended for prediction of the subway settlement.
As most railways, roads and especially metro tunnels (below the foundations of surface structures and near major urban facilities) pass through shallow urban areas, their path is not more changeable in order to achieve a reliable ground. The importance of proper analysis, support, construction method and instrumentation is revealed in this case (Palmstrom and Stille, 2006). Construction of shallow tunnels under the crowded urban areas is faced with obstacles such as settlement. Prediction, prevention and protection should be executed in order to control the settlement. Methods of prevention and controlling settlement are completely dependent on estimation. In addition, one of the most important factors in selecting the best type of support system is the amount of settlement caused by tunnel excavation. As tunnel settlement estimation depends on several parameters and each parameter could affect the others, application of the empirical or analytical methods could become impossible. To overcome these restrictions, intelligent method techniques can be used to develop a more accurate method.
In this paper, an intelligent method widely used in solving complex engineering problems, gene expression programming (GEP), has been utilized to propose new model for prediction of tunneling induced settlement. In recent years, intelligent methods have been widely used in field of tunnel engineering (Shi et al., 1998; Suihui and Guorong, 2001; Neaupane and Adhikari, 2006; Suwansawat, 2006; Suwansawat and Einstein, 2006; Santos and Tarci_sio, 2008; Teskouras et al., 2010). Based on these experiences it could be mentioned that artificial intelligence methods are useful approaches for this purpose.