Enhancing the dilute acid hydrolysis process using a machine learning approach: investigation of different biomass feedstocks influences glucose and ethanol yields
Nithianantharaj Vinitha,Jaikumar Vasudevan,K. P. Gopinath,J. Arun,S. Madhu,S. Naveen
DOI: https://doi.org/10.1007/s13399-024-05714-y
IF: 4.05
2024-05-08
Biomass Conversion and Biorefinery
Abstract:The study focuses on enhancing the efficiency of bioethanol production as a means to reduce reliance on crude oil and mitigate environmental pollution. A key aspect involves the optimization of bioethanol production through the application of an artificial intelligence approach. The Artificial Intelligence Decision-Making System (AIDMS) algorithm was developed using a machine learning algorithm, utilizing datasets derived from experimental results and published research. In the optimization process, a Pearson correlation coefficient matrix was established for 250 training datasets, revealing positive and negative correlation coefficient values. These values underscore the significance of each parameter in ethanol production. Various biomass feedstocks, including cotton stalk, wheat straw, olive tree, potato peel waste, rice straw, and sugarcane bagasse, were selected for validation of the AIDMS algorithm. The validation process compared experimental results with predictions made by the AIDMS, demonstrating a commendable 94% accuracy. The weighted rank order aggregate analysis revealed that cellulose (%), s-temp (°C), acid conc. (%), lignin (%), s-time (min), and hemicellulose (%) show the importance of parameters in obtaining glucose yield. Similarly for ethanol yield, cellulose (%), f-temp (°C), f-time (h), lignin (%), and hemicellulose (%) show the order of rank and its importance. The artificial intelligence-based optimization method is suitable for bioethanol production.
energy & fuels,engineering, chemical