Abstract

Identification of liquids is essential in chemical analysis, safety, environmental protection, quality control, and research. A novel liquid identification system based on Microwave Photonics (MWP) measured time transient evaporation signals is investigated. An extrinsic Fabry-Perot Interferometer (EFPI) based optical probe using single-mode fiber (SMF) is proposed to monitor evaporation of different liquids. The MWP system is used to measure the optical path changes during liquid evaporation due to its high sensitivity, selectivity, and Signal-to-Noise Ratio (SNR). The measured S21 continuous wave (CW) time Magnitude and Phase signals were processed to extract features such as histogram and Fast Fourier Transform (FFT) peaks. Using features extracted from droplet evaporation time transient events, machine learning classification accurately identified chemicals in each liquid with an accuracy rate of over 99%, employing three algorithms: Decision Trees, Support Vector Machine (SVM), and K-nearest neighbors (KNN). The classification results demonstrate accurate liquid identification based on evaporation measurements by the MWP system.