Calculation of filtration characteristics of wells uranium deposits

Abstract

Possession of information about the filtration characteristics of the host rocks allows you to plan the volume of ore production. The existing methodology for calculating the filtration properties of wells for uranium extraction by the method of underground borehole leaching is based on a system of rules that take into account only one geophysical parameter (apparent resistance-CS). The method gives a relatively low accuracy of calculating the filtration coefficient. In addition, it is not applicable in the case of defects in the recording of CS or distortion of values under the action of acid, which is widely used in uranium mining. At the same time, other geophysical parameters can be used to calculate the filtration coefficient, the use of which can increase the accuracy of the calculation by 10-20% when using machine learning methods.

The paper assumes the study of the applicability of machine learning methods for predicting the filtration properties of the host rocks of wells for uranium extraction by the PSV (underground well leaching) method and a numerical assessment of the advantages of such an approach. A machine learning method will be developed to predict the filtration properties of rocks based on inaccurately marked electrical logging data.