Data-Driven Multi-Failure Degradation Modeling for Neutron Generators in Logging- While-Drilling Service

This article introduces an innovative, data-driven strategy for modeling the deterioration of the neutron generator component within logging-while-drilling tools. The research commences by identifying the initial failure modes of the neutron generator and creating a health indicator (HI) to quantify the component's health status. The derived HI can be employed for further analysis and decision-making. Subsequently, a decision tree classifier is trained to establish the connection between the obtained HI values and the corresponding degradation level labels. The proposed approach is verified using real data obtained from oil well drilling operations. The experimental findings affirm its efficacy in precisely categorizing the health condition of the neutron generator component. This study is part of a prolonged initiative aimed at developing a digital fleet management system for drilling tools.

Keywords: Prognostics, Health state estimation, Oil and gas, Nuclear subsystems.