Detection of Incipient Bit Bounce Fault for Drilling Processes using Kullback-Leibler Divergence and Adaptive Alarm Deadband

Fault detection in geological drilling operations is significant for ensuring safety and reducing economic losses. To address the challenges of environmental noise interference and subtle signal variations during early-stage faults, a novel fault detection method is proposed for identifying incipient drill bit bounce. The main contributions are twofold: 1) An incipient fault detection index for drilling bit bounce is proposed based on Kullback-Leibler Divergence (KLD) between Gaussian Distributions (GDs); 2) an adaptive alarm deadband design procedure is developed to adapt to the drilling data distribution caused by the change in downhole formation. To illustrate the effectiveness of the proposed approach, an industrial case study from an actual geological drilling process is conducted.