Estimation of Shadan Gold Deposit Lithology Based on Wells Coordinate Data Using Artificial Neural Network Method.
Machine learning today has become a more effective instrument to solve many particular problems where there are difficulties problems to predicted lithology. In other words, it is a great tool for describing non-linear phenomena. We tried to use this technique to improve the existing process of predicted lithology and reduce costs on site by applying computer leaded predictions on the basis of existing on-field collected data. The dissertation describes the usage of machine learning algorithms for predicted lithology modelling based on exploration wells coordinate data for Shadan Gold Deposit. We use core analysis and well testing to determine the lithology. Unfortunately, coring from each well in the Shadan gold deposit is very expensive. However, because of the importance of this information which is obtained from lithology, it is necessary to coring from some of the deposit wells. The purpose of this study is to give a prediction of lithology in the Shadan gold deposit using an artificial neural network with back propagation algorithm (BP) and Trainlm algorithm with Mat lab software from Exploration wells coordinate data. This method can reduce the requirement of coring and reduce the costs. The area we have studied consists of six lithologies, including Andesite, Granodiorite, propylitic Alteration, Silicated Andesite, Limestone and Silica Streak. The regression between the predicted and the real values of Train, Test and Validation are obtained, respectively, as 0.94, 0.93 and 0.93. And also, RMSE the Train, Test and Validation are obtained respectively, as 0.1000, 0.1252 and 0.1012. The results show that the neural network gives a reasonable estimation for lithology.