A New Method for Joint Set Classification Based on Bayesian Optimum Classifier
By: B. Tokhmechi*, H. Memarian*, H. Ahmadi Noubari** & B. Moshiri**
Received: 2007 April 09 Accepted: 2008 May 31
Joint study is one of the primary jobs in many geological, mining, geotechnical and petroleum exploration projects. Up to 10 features of joints are gathered during each field survey, while only two of them (dip and dip direction) are normally used to classify these complex features. This paper proposes a new method for joint set classification which can use more than two surveyed features. A synthetic set of 8 joint set, each joint defined with 4 features (dip, dip direction, type of infilling and amount of infilling), created in a way that with two features (dip and dip direction) sets could not be differentiated. Necessary program developed to use Bayesian classifier to sort 8 synthetic joint sets in 4D space. Present study showed that all 8 sets can be successfully differentiated by using Bayesian method.
Kaywords: Joint, Classification, Clustering, Features, Bayesian
* School of Mining Engineering, University of Tehran, Tehran, Iran
** School of Electrical & Computer Engineering, Control and Intelligent Processing Center of Excellences, University of Tehran, Iran