Optimization of biomethane production from sweet sorghum bagasse using artificial neural networks combined with particle swarm algorithm

• Published in: Environmental science and pollution research international Vol. 30; no. 53; pp. 114095 - 114110
• Main Authors: Machineni, Lakshmi, Anupoju, Gangagni Rao
• Format: Journal Article
• Language: English
• Published: Berlin/Heidelberg Springer Berlin Heidelberg 01.11.2023; Springer Nature B.V
• Subjects: Algorithms; Anaerobic digestion; Anaerobiosis; Aquatic Pollution; Artificial neural networks; Atmospheric Protection/Air Quality Control/Air Pollution; Bagasse; Biofuels; Biogas; Bioreactors; Biorefineries; biorefining; Earth and Environmental Science; Ecotoxicology; Environment; Environmental Chemistry; Environmental Health; feedstocks; Fossil fuels; Inoculum; Lag time; Mathematical models; Metagenomics; Methane; Model accuracy; Neural networks; Neural Networks, Computer; Parameter identification; Particle swarm optimization; pollution; Process parameters; Research Article; Residues; RNA, Ribosomal, 16S; rRNA 16S; Sorghum; Substrates; sweet sorghum; Waste Water Technology; Water Management; Water Pollution Control; yield forecasting; Biomethane; Renewable energy; Machine learning; Operational parameters; Substrate to inoculum ratio; Lignocellulosic biomass; Artificial intelligence; Anaerobic digestion
• ISSN: 1614-7499; 0944-1344; 1614-7499
• DOI: 10.1007/s11356-023-30451-6

In the face of international movement away from fossil fuels caused pollution menace, many research labs are rushing towards next big breakthrough via effective biorefinery development employing non-edible agro-residues as feedstock. This work aims to evaluate and optimize the methane potential of underutilized full strength sweet sorghum bagasse (SSB) via anaerobic digestion (AD). Biochemical methane potential assays are set up for SSB AD under mesophilic and thermophilic conditions at four substrate-o-inoculum ratios (SIR) 3, 5, 7, and 9 with pH 6.5, 7.5, and 8.5 and with 80, 90, and 100 rpm mixing speed over 50 days. SIR 5 produced the highest cumulative biomethane yield of 4.25 L methane g −1 VS with a shorter lag time of 7.5 days and technical digestion time of 24 days. The influence of physio-chemical parameters on AD process dynamics is supported with 16s rRNA metagenomic sequencing. Based on complete experimental data sets, two artificial neural network (ANN) models are developed to identify the relevant significance of process parameters and to predict bagasse methane potential. Further, the developed ANN model is integrated with particle swarm optimization algorithm to create ideal AD process operating conditions which maximize the target variable, biomethane. The trained and cross-validated ANN-PSO model showed good-fit-accuracy with R 2 > 0.995 and demonstrated satisfactory performance in the biomethane yield prediction from AD of non-edible agri-residues.