This thesis explores and evaluates techniques for utilizing vibration analysis and Convolutional Neural Networks (CNNs) to assess the condition of drill bits installed on electric motor-driven drills. By strategically positioning an Inertial Measurement Unit (IMU) sensor to capture acceleration data, a wide range of vibration signals can be gathered in different operational scenarios. The CNN models undergo training and validation utilizing this data to precisely detect various fault conditions and operational states of the drill bits, showcasing the possibility of implementing scalable and reliable fault detection systems in industrial environments. The research attains a Technology Readiness Level (TRL) of 3, as demonstrated by trials that effectively categorize machine conditions using CNNs, hence confirming the critical functions of the proposed technology. The aim of this study is to assess the efficacy of vibration analysis in classifying the operational state of a drilling machine as either good, moderate, or bad.
Vibration analysis is a method used to analyze the oscillation patterns of a machine in order to identify problems such as misalignment, imbalance, and wear.