Prediction of syngas production in the gasification process of biomass employing adaptive neuro-fuzzy inference system along with meta-heuristic algorithms

Abstract Compared to fossil fuels, biomass fuels have minimal sulfur content, lower ash production, and significantly reduced emissions. The global need to reduce dependence on imported energy sources and preserve dwindling fossil fuel reserves underscores the importance of utilizing alternative energy resources. Biomass, with its abundant availability, presents a promising source for syngas production, even though the gasification procedure requires substantial energy due to its endothermic nature. Challenges related to the efficiency of biomass gasification and compliance with environmental standards have hindered economic viability. Much attention has been focused on predictive modeling of biomass gasification procedures to address these issues, necessitating robust frameworks capable of predicting parameters under varying operating conditions. This article introduces two hybrid frameworks, which are combined versions of Adaptive Neuro-Fuzzy Inference System (ANFIS) with Artificial Rabbits Optimization (ARO) and Crystal Structure Algorithm (CSA), based on proximate biomass values to predict elemental compositions (N 2 and H 2 ). These intelligent hybrid frameworks, trained with 70 % of biomass data, were further validated and tested with the remaining 15 % portions of the database. The frameworks were assessed based on some known performance metrics, namely, root mean squared error (RMSE), mean absolute error (MAE), coefficient of determination ( R 2 ), RMSE-observations standard deviation ratio (RSR), and Nash-Sutcliffe Efficiency (NSE). Developed single and two hybrid frameworks compared and obtained outcomes revealed that both introduced optimizers efficiently promoted N 2 and H 2 estimation by ANFIS, especially CSA. R 2 values for ANCS were a maximum of 0.993 in both targets’ predictions. Also, minimum RMSE values of 1.007 and 1.470 related to N 2 and H 2 prediction emphasized the accuracy of ANCS, which is capable of being used in real-world applications.