Drilling is one of the most important stages of mining operation, so, optimizing the drilling operations plays a very effective role in reducing the mining costs. In this research, the behavior measurement of the drilling process with specific energy and penetration rate has been addressed by acoustic and vibration signals produced during drilling. After the drilling tests, the acoustic and vibration signals generated during the operation were processed and the relation between the signals with the specific energy and penetration rate was eva‎luated. With the start of drilling operations on a laboratory scale, the energy consumption and penetration rate and the behavior of signals were continuously measured and investigated on each rock sample. According to the amount of drilling energy consumption and the penetration rate, three drilling stages were identified during the lifetime of drill bit. The change process of acoustic and vibration signals was similar to the change process of penetration rate and specific energy. The steps identified in the drilling mode were analyzed according to the trend of energy consumption and penetration rate using machine learning methods and statistical eva‎luations. In this case, among the different methods, the artificial neural network (ANN) method had the highest accuracy with an accuracy of 0.895. The results confirmed the accuracy and efficiency of the stages defined for the drilling mode. After this stage, by processing the acoustic and vibration signals in different drilling situations and the existence of a close relationship between the acoustic and vibration signals with specific energy and drilling penetration rate, statistical eva‎luations and development of intelligent models are carried out with the aim of classifying the drilling mode and predicting the values of specific energy and penetration rate. After running the models and testing and validating on the test data, it was observed that the random forest (RF) method with an accuracy of 0.836 has the best result in the classification of the drilling mode using acoustic and vibration signals. In the predicting the values of specific energy and penetration rate of drilling using the parameters of acoustic and vibration signals released during drilling, different machine learning models were used and the results were compared with each other. By comparing the results of various methods, it was observed that the ANN method with R2 value of 0.99 showed the best result in predicting the amount of specific energy of the drilling process. In the rock penetration rate prediction section, the investigations showed that the decision tree and random forest models with R2 value of 0.82 have the best results in predicting the value of the penetration rate of the drilling process.

راحب باقرپور

مجتبي ياري ملك آباد

مرتضي احمدي , عليرضا ياراحمدي , لهراسب فرامرزي , ابراهيم قاسمي ورنوسفادراني