Association between Rock Properties and Drill Bit Wear Based on Improved Genetic Algorithm and Deep Learning

This article aims to explore the association between rock material properties and drill bit wear through the improved genetic algorithm and the deep learning model. Firstly, the data on the physical and mechanical properties of rock samples are collected. Then, the convolutional neural network (CNN) algorithm model is used to extract its features, and based on this, a relationship model between rock properties and drill bit wear is established. Meanwhile, in this article, the genetic algorithm is optimized to enhance its efficiency and precision in solving nonlinear problems. During the experimental phase, its performance is validated through three experiments. Firstly, in the model prediction precision test, the mean squared error (MSE) of the improved algorithm is 8.54, and the mean absolute error (MAE) is 2.67. Secondly, in robustness tests under different geological conditions, the MSE of granite, limestone, and shale samples are 9.12, 8.45, and 8.88, respectively. Finally, in the model parameter optimization testing, the MSE of the improved genetic algorithm is 53.19, and the MAE is 6.42. Based on the above data conclusions, it can be found that the proposed method has good potential for application in predicting drill bit wear.