عنوان مقاله [English]
Given the importance of the rock mass rating classification system in rock engineering, the aim of this paper is to improve final classes of this classification system using k-means and fuzzy c-means clustering algorithms. The data classification in the rock mass rating classification system were allocated to certain classes via a set of initial information based on the opinions and judgments of experience, which the use of clustering algorithms in this system of classification, dataset were divided into specific classes after going through the stages of clustering analysis, therefore resulting in clarification of the final rock mass rating classification systems and removal of uncertainties from the linguistic criteria. Silhouette coefficient (SC) method was used for validation k-means clustering algorithm. Furthermore, for validation of FCM clustering algorithm, four validation methods including partition distribution coefficient (PC), clustering entropy (CE), Fukuyama and Sugeno (FS) and Xie and Beni index (XB) were used. It becomes clear that due to uncertainty condition on determination of rock mass rating clustering system classes, FCM clustering algorithms yields better results than k-means clustering algorithm. Results of data extracted from Anomaly B of Sangan iron mines indicated that the technique used in this paper is of high importance in rock mass quality.