Prediction of rock mass quality index in the Q-system by multivariate regression using the most influential parameters

Abstract

The extensive use of underground spaces is an index of development in countries. Choosing the right supporting system to achieve a safe and stable space is an important issue in tunnel construction. One of the methods used to classify rock masses is the Q-system. The Q-system depends on rock quality designation, joint strength, joint roughness, joint alteration, groundwater, and stress reduction factor, which are not always available. Sometimes it is not possible to access all the parameters of the Q-system due to time and cost. This paper aims to obtain the value of the rock quality index in the Q system using the most important parameters affecting it. Therefore, using the Pearson analysis method and SPSS software, the Q-system's most effective parameters are identified. In this regard, three models were selected to determine Q. The first and second models have three input parameters and one output parameter, and the third model has four input parameters and one output parameter. Using multivariate regression, a relationship to predict the Q-value using the most effective parameters is proposed. For this purpose, 140 experimental data were used, and 34 test data checked the results' accuracy. Determining the Q-value using three or four parameters instead of the six most effective parameters is the novelty of this paper. Comparing the results of the proposed relationship and the actual values obtained from the field measurements show that these results are in good agreement with each other. The results show that the second model with a correlation coefficient of 0.81 for the initial data and 0.8 for the test data and the root mean square error of 2.68 for the initial data and 2.55 for the test data has the best performance.