A comprehensive review on deep learning applications in advancing biodiesel feedstock selection and production processes

Abstract

Biodiesel as a renewable alternative to conventional diesel is a growing topic of interest due to its potential environmental benefits. It is typically produced from oilseed crops such as soybean, rapeseed, palm oil, or animal fats. However, its sustainability is debated, primarily because of the reliance on edible oil feedstocks and associated economic and environmental concerns. This study explores alternative, non-edible feedstocks, such as algae and jatropha, that do not compete with food production, offering increased sustainability. Despite their potential, these feedstocks are hindered by high production costs. To address these challenges, innovative approaches in feedstock assessment are imperative for ensuring the long-term viability of biodiesel as an alternative fuel. This review examines explicitly the application of deep learning techniques in selecting and evaluating biodiesel feedstocks. It focuses on their production processes and the chemical and physical properties that impact biodiesel quality. Our comprehensive analysis demonstrates that ANNs provide significant insights into the feedstock assessment process, emerging as a potent tool for identifying new correlations within complex datasets. By leveraging this capability, ANNs can significantly advance biodiesel research, producing more sustainable and efficient feedstock production. The study concludes by highlighting the substantial potential of ANN modeling in contributing to renewable energy strategies and expanding biodiesel research, underscoring its vital role in accelerating the development of biodiesel as a sustainable fuel alternative.