geochemical exploration numerical modeling using convolutional neural network (case study: gonabad region)

Modeling of mineral potentials to identify promising districts in large exploration regions for detailed exploration operations is one of the main stages of exploration. in this research, a new approach based on a convolutional neural network is proposed for geochemical numerical modeling and mineral potential exploration. in the first step, in order to create the intelligent geochemical exploration modeling, the codes of the convolutional neural network algorithm and its evaluation indicators are programmed in matlab environment. after preprocessing of stream sediment geochemical data, including identification of outliers, estimation of censored data, and data normalization and standardization, factor analysis is performed in order to reduce the dimension of the study space, identify the main variables that control the concentration of deposit elements, and define factors. the variables used in modeling are the result of factor analysis of stream sediment data. the average accuracy of the mentioned modeling is obtained as 96%. in the second step, using the geostatistical method (universal kriging), the average accuracy of estimation points via arcgis software is calculated to be 75%. at the end of this study, the performances of numerical modeling using convolutional neural network and universal kriging as well as the support vector machine and its integration with the continuous genetic algorithm, which was studied in the previous article, are compared. the evaluation results show that machine learning algorithms are more accurate in identifying promising mineral districts compared to traditional methods. it is important to note that the results of this study are in good agreement with the results of field studies and mineralized sampling.

Geochemical exploration numerical modeling using convolutional neural network (Case study: Gonabad region)

Modeling of mineral potentials to identify promising districts in large exploration regions for detailed exploration operations is one of the main stages of exploration. In this research, a new approach based on a convolutional neural network is proposed for geochemical numerical modeling and mineral potential exploration. In the first step, in order to create the intelligent geochemical exploration modeling, the codes of the convolutional neural network algorithm and its evaluation indicators are programmed in MATLAB environment. After preprocessing of stream sediment geochemical data, including identification of outliers, estimation of censored data, and data normalization and standardization, factor analysis is performed in order to reduce the dimension of the study space, identify the main variables that control the concentration of deposit elements, and define factors. The variables used in modeling are the result of factor analysis of stream sediment data. The average accuracy of the mentioned modeling is obtained as 96%. In the second step, using the geostatistical method (universal kriging), the average accuracy of estimation points via ArcGIS software is calculated to be 75%. At the end of this study, the performances of numerical modeling using convolutional neural network and universal kriging as well as the support vector machine and its integration with the continuous genetic algorithm, which was studied in the previous article, are compared. The evaluation results show that machine learning algorithms are more accurate in identifying promising mineral districts compared to traditional methods. It is important to note that the results of this study are in good agreement with the results of field studies and mineralized sampling.