A Data-Driven Approach to Detect Air Leakage in the System Using IoT Sensor Data

Compressed air is widely used as a reliable means of transporting energy in a variety of industries. However, air leaks in compressed air systems cause significant inefficiencies, resulting in significant financial losses and harm to the environment. A common solution is for a specialist to explore the site with equipment to find out if and where the leak is located. However, this solution has the limitation of requiring pneumatic equipment to be shut down at the time of the meter reading, resulting in reduced productivity and increased costs. For these reason, a lot of research is being done to monitor compressed air pressure using sensors to find leaks. In this paper, we propose a machine learning-based approach that combines DTW-based K-means clustering with the XGBoost model. The proposed method is validated by simulation using data collected from the demonstration site, and achieved an accuracy of 78.3%, AUC of 0.884, and F1 score of 0.775.