Analysis and prediction of microbial fuel cell behaviour using MLP and SVR

Background: Microbial fuel cell (MFC) is a device for simultaneous wastewater treatment and clean energy production. In this study, different amounts of yeast extract (1, 2, 3, 4, or 5 g/L) were varied and the MFC performance was measured in terms of power generation, COD removal and coulombic efficiency.Methods: The first part of this study examined a dual chamber MFC with varying volumes of yeast extract for power output, COD removal, and coulombic efficiency. Soft computing approaches were utilized in the second part of the study to estimate MFC performance, and multilayer perceptron (MLP) with varied numbers of hidden layers was used to improve model accuracy. The method was next implemented in MATLAB software using 70% and 30% of the dataset for training and testing purposes of the system, which was validated with the traditional support vector regression (SVR) estimation method.Findings: The experimental results have shown that the MFC4 (4 g/L yeast extract) had the highest catalytic activity and produced the maximum power (308 mW/m2). Comparing the experimental results and soft computing model proved that MFC behavior can be predicted 5.1819 times more accurately using MLP compared with traditional SVR.

Automated summary

In this study, the performance of microbial fuel cells (MFCs) was investigated using different amounts of yeast extract. The experiment showed that MFC4, which had a yeast extract concentration of 4 g/L, exhibited the highest catalytic activity and produced the maximum power. The study also demonstrated that using multilayer perceptron (MLP) improved the accuracy of predicting MFC behavior compared to traditional support vector regression (SVR) methods.