Machine Learning-Based Optimization of Biofuel Production from Palm Kernel Shell for Sustainable Renewable Energy Systems

Abstract

The world is moving toward a low carbon and renewable energy economy with an ever increasing demand for sustainable biofuels derived from waste from agriculture. Palm kernel shell (PKS) is one of the main wastes of the palm oil industry. Considering its high content of lignocellulose, ease of availability, and potential to reduce the environmental load from disposal then PKS has been emerged as an attractive raw material for biofuel production. Despite these, process parameters such as temperature, residence time, catalyst load and others still will have a substantial influence on the yield, energy balance as well as emissions of the PKS conversion to high quality biofuels, which makes the process inefficient. This paper offers a synthesis review of the current state of the art on the application of machine learning (ML) techniques for process optimization in context of developing biofuels based on PKS in a sustainable renewable energy (SRE) system, in which the applications are not linear but rather interdependent. It examines the combination of multi-objective optimization techniques (such as RSM, GA, PSO) with ML algorithms like neural networks, support vector machines and random forests. As illustrated in this review of the research works in the past, ML models have proven to be excellent in terms of accuracy of prediction, rapid parameter optimization for process optimization, and finally adaptive control for biofuel production systems. Besides the minimised energy input, maximum fuel output by ML + evolutionary algorithm hybrid models is even greater than the conventional ML models, and the emission control. However, data availability as well as model interpretability pose a challenge towards its application at an industrial scale. Future activities need to focus on explainable AI coupled with digital twin systems, and the use of renewable energy in processing towards highly intelligent circular carbon neutral biofuel production processing frameworks. This synthesis showcases that this transformation potential exists in the sustainability and efficiency improvements that have been achieved through ML-based optimization of PKS-derived biofuels.