Weight Determination of Mango (Mangifera indica L.) by Frequency Analysis

Abstract

The study involved the fabrication and formulation of an alternative system that use natural frequency analysis for weight measurement of mango fruits. The system was mainly composed of an ADXL345 accelerometer, Arduino UNO board, stainless steel cantilever beam, frames and breadboards, and two developed program algorithms.

The Frequency Detector (FD) and Frequency and Weight Detector (FWD) algorithms were written using National Instruments LabVIEW 2014. Thirty (30) carabao mango fruits of varying sizes were used during testing. The actual weights of the samples were obtained using a digital weighing scale. Size classification of each mango fruits based on actual weight and based on the weights predicted by the system involving natural frequency analysis was performed. The percent error using natural frequency analysis was 2.89%. Moreover, the accuracy of natural frequency analysis in size classification was determined and found that 29 out of the 30 samples were correctly classified. A verification run was performed to check the similarity of the predicted and detected frequencies of the new set of fruit samples. A t-test analysis showed that the predicted and detected frequencies have no significant difference. The average percent error between the predicted and detected frequencies was computed to be 1.10%.  The verification run established that the frequency analysis can be used to estimate weight of mango fruits and classify them in size.