An Intelligent Food Salt Tester Using Cross-Conductive Sensor

Abstract

Salt is a mineral used in various industries/foods to fulfil different needs. Its utilization for food preparation is most common and its unmonitored consumption leads to many diseases. In this article, two sensors S1 and S2 have been designed and simulated using COMSOL MULTIPHYSICS 5.6 software to measure the salt concentration of various salt samples. The performances of the sensor structure have been evaluated by varying geometrical parameters and electrode materials. The sensor works on the extension of Thompson and Lampard Theorem, a well-known principle used to fabricate a primary standard of capacitor. Sensor S2 is fabricated, and experiments are conducted with various concentrations of salt samples. The experimental results closely match the results of the simulation and the commercial conductivity meter. A highly linear response with a sensitivity of 1.87 mV/(mg/100 ml) and with an average repeatability index of ±0.17% is observed. Furthermore, the Gaussian Naïve Bayes-based machine learning algorithm has been used to categorize salt samples into eight different salt taste types. The data for salt concentration samples are generated using 100 volunteers. The machine learning algorithm achieved an accuracy of 93.33% for predicting salt taste type.

Keywords

• Circular structure
• classification of salt samples
• conductivity
• four electrode
• machine learning
• salt taster