The automation of lithology identification based on natural gamma, resistivity, neutron density and other well logging data is an important step to perform intelligent drilling/geo-steering. The current identification of lithology normally is based on the statistical results of the previous well logging data or on empirical methods, which may not be efficient or accurate. Therefore, a machine learning method is introduced here to improve the efficiency and accuracy of lithologic identification. With the development of a classification algorithm, the ensemble learning method becomes more influential since it can compensate the weak learning algorithm by using multiple learning algorithms to obtain better performance. The present research tries to identify different strata in complex sedimentary environment underground during the drilling process with a typical integrated learning method, the Adaboost algorithm, based on three wells in the Daan section, Longqian area of China. Typical single classification algorithms are used to identify the lithology, such as decision trees, SVMs (Support Vector Machines), and Bayes, etc. Comparing the results of single classifiers, the results of ensemble learning algorithm performed better than the selected single classifier. As such, the accuracy rate of lithology prediction can be increased from 66% to 90%.

You can access this article if you purchase or spend a download.