Prediction of drill performance by physico mechanical properties - An intelligent approach

Abstract

Drilling is a prerequisite operation in mining, oil and construction industry. The choice of drilling method depends on rock properties, specific production requirements, equipment availability, size and shape of the tools, nature of forces etc. Drillability of rock is the useful guide for evaluating the suitability of drills for different ground conditions. In the present study, artificial neural network technique has been used to predict the drill performance, taking physico mechanical properties, drilling parameters and drilling time as input parameters. Energy consumed and depths drilled, in given a time, are taken as measure of the performance of the drill. Two separate neural networks are designed for both the performance parameters. Each network is trained with 1500 training epochs, taking 135 and 15 data sets as training and testing data sets, respectively.