Ali Ranjbar

The high cost of drilling operations has led to increasing challenges in optimizing drilling operations. The key to success in reducing these costs is designing the well program based on the prediction of potential drilling issues and problems. Over the past few decades, the drilling industry has shown an increasing interest in machine learning to predict drilling problems. This article presents a comprehensive review of studies related to the application of machine learning in predicting high-risk drilling events. In each study, machine learning algorithms, the number of data points, input and output parameters to the machine and the performance of the corresponding algorithm are extracted from previous studies. In addition, limitations, similarities of the studies in each category are summarized and a review of the literature is provided along with recommendations for the development of future studies. These reviews show that the artificial neural network algorithm is the most popular method among the machine learning algorithms in the studies; Meanwhile, other algorithms such as support vector machine algorithm and random forest may show better performance in extracting results. It should also be noted that many of the intelligent models presented by researchers are based on limited samples and may not be generalizable to the specific conditions of presenting the results of such studies.