Abstract Geophysical well logging plays a crucial role in petroleum and natural gas exploration and production. In recent years, artificial intelligence—particularly machine learning—has driven remarkable progress in developing new logging methods and enhancing log data processing. However, research focused on logging tools themselves remains limited, primarily due to the constrained computing capabilities of downhole processors. This study reviews the current applications of machine learning in logging tools and embedded processors, examining potential implementations from three perspectives: the final log data produced by the tools, intermediate electronic signals, and other related data such as experimental measurements and noise information. Representative application scenarios include real-time prediction of logging parameters during drilling, adaptive control of acoustic excitation systems, and intelligent fault diagnosis of downhole electronic circuits. Finally, the challenges and future research directions are summarized in three key aspects: database construction, neural network model optimisation, and embedded processor selection. This review systematically sorts out the research gap of machine learning application in downhole logging tool hardware deployment, and clarifies the core technical logic of matching intelligent algorithms with embedded processors for the field. This study provides a comprehensive reference for integrating intelligent algorithms into downhole logging technologies.
Zhang et al. (Wed,) studied this question.