Correlation Study between the Organic Compounds and Ripening Stages of Oil Palm Fruitlets Based on the Raman Spectra

Area of Science:

• Agricultural Science
• Spectroscopy
• Machine Learning

Background:

• Oil palm fresh fruit bunch (FFB) maturity significantly impacts oil extraction rate (OER).
• Current methods for assessing maturity may lack speed, intrusiveness, or accuracy.
• Optimizing harvest timing is crucial for maximizing oil yield.

Purpose of the Study:

• To explore the potential of Raman spectroscopy for non-intrusive oil palm fruitlet maturity assessment.
• To develop and evaluate a machine learning algorithm for classifying fruit ripeness based on spectral data.

Main Methods:

• Collected Raman spectra from 47 oil palm fruitlets across three ripeness levels (under ripe, ripe, over ripe).
• Developed a ripeness classification algorithm using machine learning, analyzing organic compound components (e.g., beta-carotene, amino acids).
• Tested 31 machine learning models to classify fruit maturity using four significant spectral features.

Main Results:

• A machine learning approach utilizing Raman spectroscopy achieved high accuracy in classifying oil palm fruitlet maturity.
• The Medium, Weighted KNN, and Trilayered Neural Network classifiers demonstrated a maximum overall accuracy of 90.9%.
• Four key spectral features were identified as significant predictors for ripeness classification.

Conclusions:

• Raman spectroscopy presents a precise and efficient method for evaluating oil palm fruitlet maturity.
• This technique can aid in optimizing harvest timing to maximize oil extraction rates.
• The developed machine learning model offers a robust solution for automated ripeness assessment.