Monitoring of hole quality in friction drilling using different machine learning techniques

Abstract

The present investigation deals with monitoring the bush and collar formation during the friction drilling (FD) process of Be-Cu alloy. During the FD process, the alloys are often prone to structural damages as well as irregular bush formation. Hence, monitoring the quality of the hole during the process becomes inevitable. The quality of the drilled hole was monitored using different machine learning (ML) techniques. The vibration signals were captured using an accelerometer sensor. The change in amplitude of the measurement data concerning each process parameter and the measured bush surface roughness values were used to validate the effectiveness of the proposed method. The results inferred that the rigidity of the hole could be differentiated between the formation of a proper and improper bush. The study opines that the decision tree method is faster and more accurate compared to the other two methods for identifying the quality of the drilled hole.