A Digital Twin Approach for Estimation of Contaminants in Lubrication Oil through Photoacoustic Sensing System

Monitoring the changes in properties of lubricant oil with different contaminants (e.g., soot particles, water, etc.) could be a reliable technique to investigate the faults that arise due to oil contamination aiding to machine maintenance. Photoacoustic sensing system has shown promises to measure such changes, however, the attempts are limited to the collection of a large number of samples for initial calibration and standardization of the process. A digital twin substituting the system may resolve these problems. In the current work, a digital twin for the photoacoustic measurements from lubricant oil has been developed through the process of: 1) designing a simulator, 2) carrying out experiments and 3) training a domain adaptation algorithm using the simulated data and testing with experimental datasets. Analysis has shown an increase in accuracy of approximately 7% for adaptation compared to the no adaptation scenario, and has established a digital twin platform for machine health analysis.