%0 Journal Article %T Prediction of Q-Value by Multi-Variable Regression and Novel Genetic Algorithm Based on the Most Influential Parameters %J Civil Engineering Infrastructures Journal %I University of Tehran %Z 2322-2093 %A Hajiazizi, Mohammad %A Taban, Mohammad Hossein %A Ghobadian, Rasol %D 2021 %\ 12/01/2021 %V 54 %N 2 %P 267-280 %! Prediction of Q-Value by Multi-Variable Regression and Novel Genetic Algorithm Based on the Most Influential Parameters %K Genetic Algorithm %K Influential Parameters %K Multi-Variable Regression %K Q-System %K Tunnel Support %R 10.22059/ceij.2020.295339.1647 %X Determination of tunnel support, required for tunnel stability and safety, is an important debate in tunnel engineering field. Q-system classification is a technique used to determine the support system of a tunnel in rock. The problem is that all the required parameters of support system are not accessible. On the other hand, such accesses are very costly and time consuming. Therefore, it is impossible to determine the Q-value in all cases. This paper identifies the most influential parameters of Q-system using SPSS program. Then, it adopts multi-variable regression (MVR) and genetic algorithm (GA) methods to propose a relation for predicting the Q-value using three influential parameters. To this end, 140 experimental data are used. To assess the obtained models, 34 new experimental data, which are not in the primary dataset, are used. The innovation of this paper is that instead of six parameters, the Q-value is determined using three parameters with the highest impact on it instead of six parameters. In this study, the MVR model, with RMSE = 2.68 and correlation coefficient = 0.81 for train data and RMSE = 2.55 and correlation coefficient = 0.80 for test data, showed better performance than GA model, with RMSE = 2.90 and correlation coefficient = 0.82 for train data and RMSE = 2.61 and correlation coefficient = 0.84 for test data. %U https://ceij.ut.ac.ir/article_82401_ad6bb045aa2124ded9e931b3c3040f64.pdf