Machine learning approaches for predicting syngas production in biomass gasification

Abstract Fossil fuel dependence causes environmental and resource issues, intensified by climate change and population growth. Renewable sources like solar, wind, and biomass are rising. Biomass contributes 10–14 % of global energy, with gasification offering stable output and useful byproducts. However, efficiency and environmental concerns challenge its economic viability, prompting the need for predictive models under diverse conditions. This study introduces a novel hybrid modeling approach that integrates Naive Bayes (NB) with two advanced metaheuristic optimization algorithms, Jellyfish Search Optimizer (JSO) and Cheetah Optimizer (CO), to enhance the prediction accuracy of elemental compositions of nitrogen and hydrogen from proximate biomass data. The proposed hybrid schemes indicate drastically improved estimation accuracy, and among these, NBCO, i.e., the Naive Bayes + Cheetah Optimizer hybrid, was the most effective. NBCO achieved the RMSE values of 1.472 and 1.955 for nitrogen and hydrogen, validating its better prediction ability. By utilizing the synergistic properties of NB with JSO and CO, this work presents a sound predictive model complementing biomass gasification modeling with a viable tool for optimizing renewable energy processes. With the enhanced accuracy of prediction of elemental composition, the proposed schemes enable better control of biomass gasification processes with increased efficiency in energy production, reduced emissions, and decreased operation costs. Accuracy in determining nitrogen and hydrogen compositions optimizes gasifier efficiency, enabling cleaner, more cost-effective energy production. The model provides a feasible solution for businesses and policymakers seeking to maximize the potential of biomass as a renewable energy source while minimizing environmental and economic problems.