查看文章

This paper presents a method for constructing a health indicator to detect neutron-generator faults in a multifunction Logging-While-Drilling (LWD) service and predict maintenance requirements due to wear. The method is based on extracting features from selected channels that hold information about the subsystem degradation with time. These features are used to build a decision-tree model which estimates the tool condition from the recorded data. The model demonstrates excellent value for both maintenance and field engineers due to the fact that in just a few minutes the physical condition of the neutron generator can be determined with high confidence. This work is part of a long-term project with the aim to construct a digital fleet management for drilling tools.