Design and applications of a neural networks assisted portable liquid surface tensiometer

Abstract

In this paper, a portable instrument for surface tension measurements, characterization and applications is described. The instrumentation is operated wirelessly, and samples can be measured in situ. The instrument has changeable different size probes; therefore, it is possible to measure samples from 1 ml up to 10 ml. The response of the measured retraction force and the concentrations of measured surfactant is complex. Therefore, two calibration methods were proposed: (i) the conditional calibration using polynomial and logarithmic fitting and (ii) the neural network trained model prediction of the surfactant concentration in samples. Calibrating the instrument, the neural network trained model showed a superior coefficient of determination (0.999), comparing it to the conditional calibration using polynomial (0.992) and logarithmic (0.991) fit equations.