Method for drill tool thread damage assessment based on APMR-GBM

This study addresses the prevalent issue of fatigue-induced fractures in tool joint threads during oil and gas drilling operations. In response to the practical limitations of existing metal magnetic memory (MMM) inspection methods—which are often susceptible to interference from material magnetization intensity, lift-off distance, and other influencing factors—a novel damage assessment approach based on the area-based peak-to-average ratio of magnetic signals is proposed. This method effectively mitigates the impact of such interferences on inspection outcomes. Through a combination of simulation analysis and field experiments, magnetic signal data on the thread surface were systematically collected, and multiple characteristic parameters were extracted. Machine learning techniques were further introduced to compare various classification models. Results indicate that the Gradient Boosting Machine demonstrated superior performance in identifying thread damage, with evaluation metrics such as accuracy and recall rates significantly outperforming other models. This research provides a new direction for automated and efficient damage inspection of drilling tool threads, offering practical value for enhancing the safety of drilling operations.