Intelligent interpretation method for lithology recognition using real-time LWD data

The lithology of the drilling formation plays an important role in the drilling performance by affecting rate of penetration (ROP), drilling risks, and engineering parameters, etc. To solve the complexity and subjectivity in lithology interpretation of logging while drilling (LWD), this paper proposes an intelligent real-time lithology interpretation method based on logging while drilling, which is used in real-time lithology recognition from logging while drilling data. To ensure the requirement of real-time efficiency, the logging parameters in the working area have been interpreted manually as the training data of the neural network, the input is the logging data measured while drilling, and the output is the interpretation result of the corresponding lithology. The neural network is trained with a group 253 logging while drilling (LWD) sample data in a certain block, and a group of data is selected to test the automatic interpretation of the neural network and its popularization and application effect. The test results show that the method has high accuracy for automatic interpretation using logging while drilling data, and the accuracy of 90.7% is obtained in the test set. The experimental data test shows that this method can automatically, efficiently, and accurately provide lithology interpretation during the drilling process. Therefore, the proposed method can be used to assist drilling engineers in real-time drilling adjustment and geo-steering, and contribute to improve the reliability and efficient in the drilling process.