Predicting the value of the rock quality index in the Q-system using gene expression programming

Among the methods used to design the tunnel, the Q-system is a comprehensive method that has attracted the attention of many researchers today. However, the limitations of the Q-system make it impossible to access all the required parameters as well as the time and cost of them,  which has made it impossible to classify the rock mass using the Q-system. This paper attempts to predict the value of Q by parameters that have the highest coefficient of importance in the value of Q, using the Gene Expression Programming (GEP) technique. The most effective parameters involved in the Q value have been identified using Pearson correlation analysis (PCA), and then three different input models have been used to obtain Q value so that they are more closely related to experimental values. A total number of 159 experimental data were used for training and testing of the models, respectively. The innovation of this paper is that instead of 6 parameters, only three influential ones were used for determining the value of Q. Using the three parameters RQD, Jn and Ja, which have been determined as the most effective parameters and applying Pearson correlation analysis method, the value of Q can be determined with an acceptable approximation. In the suggested relation, the coefficients of determination (R2), root mean square error (RMSE), BIAS and the scatter index (SI) obtained were 0.917, 2.31, 1.74 and 0.43, respectively that show the new equation presented by GEP, can be undoubtedly used to predict the value of Q.