Identification of Liquid Substances of Similar Chemical Composition by the Acoustic Method

A comprehensive approach to processing the responses of Lamb acoustic waves propagating in piezoelectric plates with a liquid coating has been developed, which allows for the differentiation of liquids with similar chemical composition and electrical conductivity. The technique is based on the registration of the amplitudes of zero-order and higher-order Lamb modes in a multi-channel sensor system and the application of polar histogram visualization and principal component analysis (PCA) methods to process the measured data arrays. The technique was tested using 33 acoustic modes for 8 samples of mineral water collected from various sources in the North Caucasus. It was found that even when the compositions are similar, the analyzed liquids exhibit differences in their histograms that are more than an order of magnitude. Changes in the responses were observed during the storage of the samples, indicating the "aging" of the substance. A comparative analysis of the results obtained using polar histograms and the standard PCA classification method was performed. The obtained results are of interest for the development of sensors that do not require sensitive coatings, as well as for integration with machine learning algorithms when working with other types of sensor data.