A Non-Destructive Oil Palm Fruit Freshness Prediction System with Artificial Neural Network

Abstract

The economical, rapid, and non-destructive method using reflectance Near-Infrared Spectroscopy (NIRS) technique were designed and developed for oil palm fruit (Elaeis Guineensis) freshness prediction. The ripe maturity oil palm freshness of Tenera variety was used for this study by two consecutive days. The measurement of spectral value was obtained with a linear array sensor. The Artificial Neural Network (ANN) was trained with Levenberg-Marquardt algorithm by using half of the data sets, quarter for validation and the rest quarter for test purpose. The performance of the oil palm fruit freshness prediction system was evaluated. Results indicate that the ANN with 6 hidden neurons achieved the best prediction accuracy with root mean squared error (RMSE) and the correlation coefficient (R) were 6.8449 hours and 0.8418 respectively. This suggests that the proposed method is promising to be further developed to automate oil palm freshness inspection.