Application of artificial neural network and particle swarm optimization for modeling and optimization of CO2 sequestration using microalgae

Department of Chemical Engineering, Department of Earth and Environmental Studies, Department of Chemistry,

National Institute of Technology Durgapur, Durgapur-713 209, West Bengal, India

E-mail: susmita.dutta@che.nitdgp.ac.in

Manuscript received online 29 May 2020, revised and accepted 25 August 2020

In the present study, Chlorococcum sp., a green alga – collected from coke-oven effluent treatment plant, was used for the sequestration of carbon dioxide at diverse initial CO₂ concentrations (5-30%), pH of the medium (7-11) and inoculum sizes (5-12.5%), respectively. The test strain showed sustainability in growth upto 30% CO₂. However, the maximum CO₂ sequestration was found to be 79.08±0.89% at 5% CO₂, pH 9 and 12.5% inoculum size for the test strain. Since phenomenological modeling is difficult to build in such systems, artificial neural network modeling technique was tried in this work to predict the CO₂ removal at dissimilar pH of the medium, diverse CO₂ concentrations and at various inoculum sizes for the test strain. Various algorithm of artificial neural network (ANN) model and lot of activation functions were tried to arrive at best ANN model for this experimental data. Developed ANN model performance was found very efficient. Optimization by particle swarm optimization (PSO) algorithm is carried out to determine the optimum value of input parameters to sequester maximum CO₂. With PSO algorithm the optimum input parameters are obtained to achieve 99% removal of CO₂. As far as known the studies on application of Artificial Neural Network and Particle Swarm Optimization for modeling and optimization of CO₂ sequestration using microalgae are very rare and here lies the novelty of the present work.