Event-Driven Explainable Fault Diagnosis Method for Drilling Pump

Accurate fault diagnosis is a critical aspect of Prognostics and Health Management (PHM) for drilling pumps. The variable speed and load characteristics of drilling pumps present challenges for extracting fault-sensitive features. In this paper, a novel event-driven explainable fault diagnosis method for drilling pumps is proposed. The influence of rotational speed on original signals is eliminated, the correlation between component motion and vibration impacts is established, and the most sensitive features representing the operational state of a drilling pump are extracted, resulting in high-precision fault diagnosis. The diagnosis process employs order tracking and angular domain resampling to convert time-domain non-stationary original signals into angular domain stationary signals. Critical motion events are determined based on the correlation analysis method for rotating-reciprocating components. A fault identification method based on a cloud model is constructed. Finally, the proposed method is validated by experiments resulting with an average accuracy of 98%, which is significantly higher than other methods.