Determination of Moisture Content from Microstrip Moisture Sensor with Minimum Mean Relative Error

Abstract

Aiming for better food security in any country, microstrip moisture sensor is designed to detect moisture content in essential food grains. While designing sensor, the motive is to have a minimum mean relative error (about to zero) so that accuracy can be achieved. Initially, triple-frequency sensor is designed and fabricated on FR4 material sheet. During study, rice is used as a sample for analysis. Performance of sensor is analyzed by determining the mean relative error, which is calculated with the help of actual moisture content and predicted moisture content. Hence, both actual moisture content and predicted moisture content are calculated. Actual moisture content, calculated with the help of microwave oven drying weight technique, varied from 10.71% to 21.87% for five samples at different water contents. Predicted moisture content, calculated with the help of vector network analyzer and calibration equation, varied from 10.69% to 21.84%. Hence, the mean relative error obtained is 0.16% for triple-frequency sensor. It is clear that the mean relative error is not close to 0. Now, moving forward, four-frequency sensor is designed, whose predicted moisture content is varied from 10.70% to 21.85% and a mean relative error is found to be 0.073%. Hence, four-frequency sensor will be more preferable because of its accuracy in comparison with triple-frequency sensor. But, any of the sensors can be adopted depending on the accuracy required. Both can be adopted in fields such as agriculture, horticulture, textile industries, and various others.