Impact of optimization parameters on green sustainable oil from ripe palm kernel seeds for energy utilities: A machine learning approach

Abstract

In this study, two predicting tools; Artificial Neural Networks, and Machine Learning model were used to optimize and predict the green sustainable oil synthesis from ripe palm kernel seeds utilizing the Soxhlet extractor with n-hexane. The Artificial Neural Networks result indicates that the highest predicted value obtained was 37.769 wt percent at 50 min of extraction time, 175 ml of solvent volume, and 40 g of sample weight, while the highest predicted value from Machine Learning approaches was 38.0354 wt percent at 40 min of extraction time, 200 ml of solvent volume, and 60 g of sample weight. Thus, Machine Learning produced higher yield. Also, the Artificial Neural Networks model's R2 was 0.99922, and the Machine Learning model's R2 was 1 which indicates the muscular nature of Machine Learning. Furthermore, the results of the sensitivity analysis from the machine learning approach based on Python instruction gave an easy decoding of the most crucial input factors and their impact on the predicted values, correlating with other findings from this study. Additionally, the physicochemical assets of green sustainable oil produced from ripe palm kernel seeds falls within acceptable range. Thus, this shows that the Obetim-Uno ripe palm kernel seeds could be used to cool and lubricate whirling machinery parts, in addition to being used to produce green-fuels, when treated with other materials.

Keywords

• Optimization
• Obetim-uno seeds oil
• Soxhlet extraction
• Green sustainable oil
• Artificial neural networks
• Machine Learning model