Application of cluster analysis on Rock Mass Rating system

• Main Authors: Yu Siu Tzeng, 曾于修
• Other Authors: C.T.Chang;C.T.Lee
• Format: Others
• Language: zh-TW
• Published: 1995
• Online Access: http://ndltd.ncl.edu.tw/handle/72503535539810547995

碩士 === 國立中央大學 === 應用地質研究所 === 83 === Rock mass Rating(RMR) System is important in deciding the types and amounts of tunnel supports. Traditionally, RMR system is used to categorize rock mass into different grades. However, rock mass with same grades do not necessarily possess similar geologic conditions. This question is of practical weakness to the RMR system. The purpose of this study is to initiate a new approach of RMR system by using principal component analysis and cluster analysis. The scope of this tudy includes 2019 and 1804 geological data recorded from excavation faces of the Futeh tunnel and Sanyih #1 tunnel respectively. The objectives of principal component analysis are filtering out data variability and undertaking the rotation of principal component axis. Cluster analysis is performed to identify the RMR samples with different weighting factors and then, to classify those samples according to their similarities. This study use Ward''s method, one of hierachical cluster analysis method, to find the error of sum of square,and then to classify samples accordingly. By using this method, mor reasonable result can be achieved. To deal with the records, we still follow the current weighting of RMR system to compare with the source system and the emperical functions. In practice, this method can help solving problems of which rating are located near classification boundary of RMR system and helps the decision of tunnel supports.