Classification of Sonic Log using Multiresolution Features

This research work proposes a novel approach to sonic log prediction in the oil and gas industry, emphasizing the importance of accurate DTC (Compressional Sonic Transit Time) and DTS (Shear Sonic Transit Time) values for reservoir characterization. The methodology combines machine learning and signal processing, specifically Wavelet transformation, to enhance prediction accuracy. The process involves dataset preprocessing, Wavelet transformation on seismic signals, feature extraction, and machine learning model training. Results demonstrate an impressive 89 percent accuracy, surpassing other algorithms. The dataset utilized is comprehensive and diverse, ensuring the model’s robustness across various geological scenarios. The proposed approach significantly contributes to accurate wave velocity estimation, boosting confidence in reservoir characterization. Overall, the study offers a practical and effective solution for improving sonic $\log$ predictions, benefiting hydrocarbon exploration and decision-making in the petroleum industry.