Selective gas detection using conductivity-based MEMS resonator and machine learning

This work demonstrates multiple gases identification using a heated MEMS resonator and machine learning. The working principle of the gas sensor is based on the cooling/heating effect of the injected gases on the electrothermally actuated micro beam. As a case study, we demonstrate the concept using two analytes: Acetone and Helium. Machine learning algorithms and Principal Component Analysis are employed to classify each gas with its specific concentration level. The results show that a 100% accuracy rate is achieved for the identification of the different analytes with their concentration levels.