本文主要在於發展一不連續面空間位態劃分組數之模式,其乃根據統計方式與叢集理論來分析五種目標函數,以估計各目標函數之極值,並獲得不連續面空間位態最佳可能劃分組數方式。同時,當獲得組數後,並以共變數法及傳統合向量法來估計每組最佳平均集中位態。為檢定每組空間分佈之合適性,以三種球體分佈函數(Fisher, Bivariate Fisher和Bingham分佈函數)及卡方試驗來檢定之。文中並與前人研究相互比較,結果顯示本文模式可包含前人之模式,且較前人之模式具統計意義。
This paper provides a statistical technique with the cluster analysis to study clusters of discontinuity orientations in rock mass. Five objective functions are used to minimize constraint on resulting partitions in order to delineate clusters of discontinuity orientations. When discontinuities are clustered, the precision mean direction of each set is obtained by using resultant vector method and co-variance matrix method. The chi-square test is used to test three orientation distributions (Fisher, Bivariate Fisher and Bingham distributions) of each set. The technique for delineation and analysis of clusters is applied to an example problem through use of a computer code. In comparison with the technique of Shanely and Matatab's method, the results show that the technique of this paper covers their processes and it is more statistieally meaningful.