Prediction of Thrust and Torque in Drilling Using Conventional and Feedforward Neural Networks

Drilling performance prediction, using traditional mechanics of cutting approach, is based on the extension of three-dimensional oblique cutting theory. The quantitative reliability of such conventional models depend on a numerous number process variables and quantitative accuracy of the data bank for a given work material. The complexity of such models is increased when inevitable eccentricity and drill deflections are incorporated into the analysis. In this paper, using a novel neural network architecture that optimises the output layer, the thrust and torque in drilling operation are carried out. A set of comprehensive drilling tests is carried out to train and test the architecture. It has been shown that the percentage deviations of drilling predictions using the neural network architecture is -0.56%, and 1.03% for thrust and torque compared to 4.20% and –10.25% using traditional mechanics of cutting approach.