Mini-Review on Petroleum Molecular Geochemistry: Opportunities with Digitalization, Machine Learning, and Artificial Intelligence

Abstract

Molecular geochemistry plays a vital role in understanding the origin of oil and gas, correlating hydrocarbons with their source rocks, and evaluating the potential of source rocks. However, traditional molecular geochemistry methods increasingly struggle to meet the demands of modern exploration due to their complexity and inefficiency. This challenge is particularly pronounced in the digital era, where petroleum exploration is characterized by continuous refinement and the growing prominence of unconventional hydrocarbons. To address these challenges, various machine-learning techniques, leveraging statistical and chemometric principles, have emerged as effective solutions. This review analyzes the application and challenges of machine-learning-based methods in molecular geochemical data processing, highlighting both unsupervised techniques (such as hierarchical cluster analysis and principal component analysis) and supervised approaches (including artificial neural networks). Additionally, it explores the future development of machine learning in petroleum molecular geochemistry, emphasizing the creation of integrated big data systems and intelligent analysis tools. This includes the use of advanced technologies, such as digitalized chromatograms and convolutional neural networks, which promise to further enhance data interpretation and decision-making in petroleum exploration.