Energy analysis and optimization of a biosensor-based microfluidic microbial fuel cell using both genetic algorithm and neural network PSO

Abstract

Since the lack of parametric analysis and optimization on the output power density was considerable in previous studies, this study tries to analyze the impact of the most critical parameters on the power density. In order to find these parameters, a conduction-based model was applied and the model was calculated by using a neural network. The neural network was trained by using a Particle Swarm Optimization (PSO). These parameters are microorganism's conductivity, external resistance, anode surface area, microchannel height, and temperature. A sensitivity analysis was employed to find which parameters can be negligible, and the results showed that temperature is not as effective as other parameters and could be neglected in the optimization. To maximize the power density, both Genetic Algorithm (GA) and Neural Network PSO (NN–PSO) was used, then the run-time of each were compared. To validate the present study, the results were compared with the results of an experimental study and a good fit was observed. The maximum power density obtained from other studies was compared to the result of both optimizations in this study, and showed 46% improvement in the power density. The values of parameters, by which the maximum power density was achieved are Rext = 1 Ω, Aan = 4.152 m2; Hmicrochannel = 100 μm; kmicroorganism = 0.001 mS cm−1. Finally, run-time of both optimization methods was compared and it is found that NN-PSO decreased the run-time significantly from 15 h to 17 min.