Artificial neural network based modelling and optimization of microalgae microbial fuel cell

Abstract

Simultaneous wastewater treatment and energy harvesting is attractive topic these days. A microbial fuel cell is an electrochemical device that can be used effectively for this purpose. Microalgae-based MFC is a novel approach to extracting sustainable and economical energy by incorporating photosynthesis with MFC. This paper uses artificial intelligence to identify the best operational factors of microalgae microbial fuel cell (MMFC). The proposed methodology integrates artificial neural network (ANN) modelling and forensic-based investigation algorithm (FBI). Yeast concentration (%) and wastewater concentration (%) are used as decision variables during the optimization process, whereas the objective function is simultaneously maximization of power density and COD removal. Based on the measured data, a ANN model is designed to simulate the power density and COD removal in terms of yeast and wastewater concentrations. Compared with ANOVA, the values of coefficient-of-determination are increased. For the power density model, the coefficient-of-determination in the prediction is increased from 0.7275 to 0.9783 by around 34%. Whereas for the COD removal model, the coefficient-of-determination in the prediction is increased from 0.8512 to 0.9 by around 5.7%. Then, using FBI, the best concentrations of yeast and wastewater are identified to increase power density and COD removal simultaneously. To prove the superiority of the proposed methodology, the optimal parameters and best performance are compared with an optimized performance by response surface methodology and measured data. The performance of MMFC is increased by 2.24%, thanks to the integration between ANN and FBI.