Estimation of the Presence of Coal Using ANN Method by Employing the Geophysical Log Parameters at the Soma Basin

Abstract

In this study, the feed forward, back propagation neural network model (ANN), used as one of the complex modeling systems and artificial intelligence methods, have been employed to estimate the presence of a coal layer at the Soma Coal Basin. Considering the costs of potential drillings, using ANN means advantages in terms of labour and cost. In this work, the neutron, gamma ray and density data are used as input parameters to the ANN system. For the study case, the geophysical well logs parameters of ten mechanically are drilled wells in the Soma Coal Basin are used. For four drilled wells, the best R and MSE values of 0.74 and 0.084 were estimated respectively. Whereas for each well separately values 0.81 and 0.051 were estimated. It is concluded that ANN given a good fit for the accurate prediction of the presence of coal layers from well log data. Furthermore, for modeling of reserves and estimation of coal layer characterization, ANN gives quick and highly accurate results.