Rock mass rating system （RMR） is based on the six parameters which was defined by Bieniawski （1989） [1]. Experts frequently relate joint and discontinuities and ground water conditions in linguistic terms with rou...Rock mass rating system （RMR） is based on the six parameters which was defined by Bieniawski （1989） [1]. Experts frequently relate joint and discontinuities and ground water conditions in linguistic terms with rough calculation. As a result, there is a sharp transition between two modules which create doubts. So, in this paper the proposed weights technique was applied for linguistic criteria. Then by using the fuzzy inference system and the multi-variable regression analysis, the accurate RMR is predicted. Before the performing of regression analysis, sensitivity analysis was applied for each of Bieniawski parameters. In this process, the best function was selected among linear, logarithmic, exponential and inverse func- tions and finally it was applied in the regression analysis for construction of a predictive equation. From the constructed regression equation the relative importance of the input parameters can also be observed. It should be noted that joint condition was identified as the most important effective parameter upon RMR. Finally, fuzzy and regression models were validated with the test datasets and it was found that the fuzzy model predicts more accurately RMR than reression models.展开更多
文摘Rock mass rating system （RMR） is based on the six parameters which was defined by Bieniawski （1989） [1]. Experts frequently relate joint and discontinuities and ground water conditions in linguistic terms with rough calculation. As a result, there is a sharp transition between two modules which create doubts. So, in this paper the proposed weights technique was applied for linguistic criteria. Then by using the fuzzy inference system and the multi-variable regression analysis, the accurate RMR is predicted. Before the performing of regression analysis, sensitivity analysis was applied for each of Bieniawski parameters. In this process, the best function was selected among linear, logarithmic, exponential and inverse func- tions and finally it was applied in the regression analysis for construction of a predictive equation. From the constructed regression equation the relative importance of the input parameters can also be observed. It should be noted that joint condition was identified as the most important effective parameter upon RMR. Finally, fuzzy and regression models were validated with the test datasets and it was found that the fuzzy model predicts more accurately RMR than reression models.
