Geochemical pattern recognition for Cu-Au Deposit Based on Self-Organizing Map (SOM) and Fuzzy K-means Clustering (FKMC) in Meshginshahr, NW of Iran
محورهای موضوعی : journal of Artificial Intelligence in Electrical Engineering
Aynur Nasseri
1
(Assistant Professor, Department of Mining Engineering, Ahar Branch, Islamic Azad University, Ahar, Iran)
کلید واژه: Meshginshahr, geochemical pattern recognition, elements distribution, Self-Organizing Maps (SOMs), Fuzzy K-means clustering (FKMC),
چکیده مقاله :
Mapping the mineralized zones and providing an appropriate distribution pattern of elements for characterizing geochemical system and targeting potentially promising areas of Cu-Au mineralization by utilizing an adequate technique and establishing an optimized exploration tool is the main object of this study in Meshginshahr, NW of Iran. In this respect 144 stream sediments samples were collected and analyzed for Au, Ba, Bi, Cd, Ce, Co, Cr, Cu, Hg, Mo, Ag, As, Sn, Sb, W and Pb. In this study, self-organizing map (SOM) and Fuzzy K-means clustering (FKMC) approaches with the aim of pattern recognition were employed. The SOM as a dimension reduction approach was introduced to recognize geochemical dispersion patterns with high certainty while preserving the originality of data.. During data processing, SOM appropriate structure with a pattern including six clusters was selected and the related elements distribution model was extracted. Results represent two significant sets of elements in clusters for anticipating the mechanism of distribution. In this target pattern, copper and pertaining trace elements formation are localized in the north of the area. Also, Au Anomalies and its associated elements are mostly elongated from NW to SW of the area. To evaluate the SOM results, a comparative study was carried out with the results obtained from Fuzzy K-means clustering (FKMC). FKMC performance showed the proper compliance with the SOM results with respect to the relationship between the elements and their corresponding membership’s probabilities in different clusters. The results illustrated higher performance of the approaches in characterizing geochemical pattern and detecting the element paragenetic sequence in the area for locating the exploration targets..
