Lithology Mapping by Affinity Propagation Algorithm
Nayara Safira da Silva Caldas, Carlos Guerra, Cleudilene Silva Rufino · 2017
This work presents a methodology for solving the problem of the identification of lithologies at depth, directly from wireline logs. For this purpose, the M-N cross-plot is used as the formation evaluation technique for identifying the lithologies in the logged borehole, in terms of the physical properties of the main mineral. However, the visual interpretation of this graphic is limited by the large spread of points in the graph. The proposed method is algorithm for a computational interpretation of the M-N data by means of the Affinity Propagation (AP) clustering technic. A minimum distance criterion, with respect to M-N fixed points, is applied to the set of exemplary points, acquired from AP, in order to associate the lithologies. Finally, the methodology was applied on a real M-N data, proving its effectiveness, both in reducing the number of groupings obtained by the clustering algorithm as well as in the lithological identification over highly-spread M-N data set.