Machine learning in sustainable fiber-reinforced polymers: a bibliometric and critical assessment

Abstract

The growing demand for sustainable materials has accelerated the development of bio-based fiber-reinforced polymers (BFRPs) as environmentally friendly alternatives to conventional composites. However, the intrinsic variability of natural fibers and the complexity of processing–structure–property relationships pose significant challenges for material design and optimization. In this context, machine learning (ML) has emerged as a powerful data-driven approach for predicting material behaviour and accelerating the development of sustainable composites. This paper presents a comprehensive bibliometric and critical review of ML applications in BFRPs. Publication trends, influential contributors, collaboration networks, and thematic evolution are analyzed to map the development of this interdisciplinary research field. The review further examines the most frequently used ML algorithms, targeted material properties, and investigated bio-fiber systems. Particular attention is given to methodological practices, data limitations, and model validation strategies. The analysis reveals a rapid increase in ML-based studies, with artificial neural networks, support vector machines, and tree-based methods dominating the literature. While promising predictive capabilities have been demonstrated, challenges related to data quality, model interpretability, and generalization remain. The study concludes with recommendations for future research directions to enhance the reliability, transparency, and sustainability impact of ML-assisted BFRP development.