Machine Learning Inspired Classification Model of Different Edible Oil Samples using High Sensitivity Microwave Sensor

Food safety has received a lot of attention over the recent years, and the quality of cooking oil is very important. Adulteration is very common in edible oils. Studies have been conducted using electromagnetic waves to analyze the properties of materials. Henceforth, the quality of edible oil can be checked by measuring the dielectric constant of the oil sample. Microwave sensors are of great interest due to their superior sensitivity, durability, and affordability. A complementary split ring resonator (CSRR) based compact sensor with high quality factor having dimensions as 15 X 25 X 0.8 mm 3 is designed on Rogers RT-Duroid 5880 substrate with relative permittivity 2.2. The designed sensor shows frequency shift with change in the permittivity of the oil samples used, showing high sensitivity. In this work, a machine learning integrated oil classification approach has also been proposed utilizing the above designed microstrip based sensor. Out of various models used, gradient boosting classifier and naive bayes showed maximum accuracy of almost 96%.