Maximizing Bio-Hydrogen Production from an Innovative Microbial Electrolysis Cell Using Artificial Intelligence

Abstract

In this research work, the best operating conditions of microbial electrolysis cells (MECs) were identified using artificial intelligence and modern optimization. MECs are innovative materials that can be used for simultaneous wastewater treatment and bio-hydrogen production. The main objective is the maximization of bio-hydrogen production during the wastewater treatment process by MECs. The suggested strategy contains two main stages: modelling and optimal parameter identification. Firstly, using adaptive neuro-Fuzzy inference system (ANFIS) modelling, an accurate model of the MES was created. Secondly, the optimal parameters of the operating conditions were determined using the jellyfish optimizer (JO). Three operating variables were studied: incubation temperature (°C), initial potential of hydrogen (pH), and influent chemical oxygen demand (COD) concentration (%). Using some measured data points, the ANFIS model was built for simulating the output of MFC considering the operating parameters. Afterward, a jellyfish optimizer was applied to determine the optimal temperature, initial pH, and influent COD concentration values. To demonstrate the accuracy of the proposed strategy, a comparison with previous approaches was conducted. For the modelling stage, compared with the response surface methodology (RSM), the coefficient of determination increased from 0.8953 using RSM to 0.963 using ANFIS, by around 7.56%. In addition, the RMSE decreased from 0.1924 (using RSM) to 0.0302 using ANFIS, whereas for the optimal parameter identification stage, the optimal values were 30.2 °C, 6.53, and 59.98 (%), respectively, for the incubation temperature, the initial potential of hydrogen (pH), and the influent COD concentration. Under this condition, the maximum rate of the hydrogen production is 1.252 m3H2/m3d. Therefore, the proposed strategy successfully increased the hydrogen production from 1.1747 m3H2/m3d to 1.253 m3H2/m3d by around 6.7% compared to RSM.