Application of Artificial Intelligence (AI) to biomass pyrolysis system

The palm oil industry generates million tonnes of oil palm biomass annually. Biomass conversion using the thermochemical pyrolysis process is interesting as it produces value-added products such as bio-oil, biochar and product gas. However, accurately predicting pyrolysis processes is challenging due to multi-scale complex reactions. Therefore, the objective of the present research is to apply AI models, such as artificial neural network (ANN), support vector machine (SVM), decision tree (DT) and particle swarm optimization (PSO) to predict the oil palm biomass pyrolysis behaviors and product yield Phase 1 research work focuses on the prediction of pyrolysis behavior of two types of oil palm biomass; empty fruit bunches (EFB) and oil palm shells (OPS) in a thermogravimetric analyzer at heating rates (5, 10, 15, and 20°C/min). The results showed that AI models successfully predicted the pyrolysis behavior of biomass, achieving high coefficients of determination (R2) ranging from 0.94 to 0.99 and low mean square errors (MSE) ranging from 0.09 to 5.12. However, product yield prediction was less accurate, with higher MSE values (4.27, 2.79, and 6.67). Phase 2 of the research work was carried out in a lab-scale tube furnace with varying temperature, particle size and carrier gas flow rate. The data obtained from experiments were used to train the hybrid ANN-PSO model. Results revealed that increasing the complexity of the dataset does not cause a major decline in the AI model's performance (R2 = 0.921, MSE = 0.0055). The application of AI in biomass pyrolysis holds promising potential for the oil palm biomass industry that can improve process optimization and enhance product yield and quality. Moreover, AI technologies contribute to job creation and human capital development in AI, bioenergy, and biofuels. Integrating AI also leads to cost savings by reducing analysis costs associated with biomass pyrolysis.