Abstract
Using AI-Based sound wave analysis of drilling equipment to assess its health condition and to predict equipment failure by detecting anomalous sound patterns in both time and frequency domains, and apply predictive maintenance.
Machines can produce noise with frequencies higher than the upper audible limit of human hearing, which is referred to as ultrasound (or ultrasonic sound). Thus, by listening to a wider sound spectrum, a better understanding of equipment state is achieved; and will lead to more accurate failure prediction methods. The process starts by capturing sounds via microphone. Then, Short Time Fourier Transform is used to convert time domain sound wave to its corresponding frequency domain. Both time and frequency domain signals are combined to form a two-dimensional spectrogram. This spectrogram is trained using an AI model to learn the difference between normal and abnormal equipment sounds. After training, the system identifies anomalous sounds the equipment might produce. Hence, generate an alert to the operational team to take actions, preventing equipment failure.
Compared to the human auditory system, the experimental results showed that the proposed method achieved significant improvement in anomalous sounds detection only for machines that can produce noise in the ultrasound region, while other machines achieved worse results compared to the audible sound region.
