Biohydrogen has attracted significant global attention as a clean energy carrier that can support the transition to low-carbon energy systems. Among renewable feedstocks derived from agricultural residues, sugarcane bagasse (SCB) presents strong potential for sustainable biohydrogen production through gasification, offering advantages in resource availability, waste valorisation, and reduced greenhouse gas emissions. However, large-scale biohydrogen deployment remains constrained by challenges related to feedstock variability, process efficiency, economic viability, supply chain integration, and downstream purification and storage. The inherent complexity and nonlinearity of biomass gasification processes further limit the accuracy of conventional modelling and optimisation approaches. Machine learning (ML) techniques have emerged as powerful tools to address these limitations by modelling complex multivariate relationships and enhancing predictive capability under variable operating conditions. These models incorporate temperature, pressure, and feedstock characteristics to enable predictive decision-making, real-time monitoring, flexible control strategies, and optimisation. This review provides a comprehensive synthesis of recent developments in modelling and predicting biohydrogen production from biomass gasification using ML algorithms, including random forests, artificial neural networks, support vector machines, and regression-based approaches. The operational mechanisms, advantages, limitations, and predictive performance of these methods are evaluated with respect to process optimisation, yield forecasting, and classification tasks. While ML models demonstrate strong capability in handling nonlinear and limited datasets, challenges such as overfitting and generalisability remain important considerations. By integrating technological, modelling, techno-economic, and sustainability perspectives, this review addresses a critical gap in linking biomass gasification research with data-driven optimisation strategies, particularly for SCB-based systems. The study provides structured insights to improve process efficiency, scalability, and industrial applicability, thereby supporting the advancement of sustainable, ML-enabled biohydrogen production aligned with global decarbonization and sustainable energy goals.
Maluleka et al. (Sun,) studied this question.