Prediction of Biological Hydrogen Production in a Packed-Bed Bioreactor Using a Genetically Evolved Artificial Neural Network

In this study, a fermentative hydrogen-producing bacterium, Clostridium tyrobutyricum JM1, was isolated from a food waste treatment process. The isolate was immobilized in a packed-bed bioreactor using polyurethane foam as a support medium. The performance of the reactor was predicted by a feed-forward backpropagation neural network (FBNN) whose structure and weights were genetically evolved using a genetic algorithm (GA). The GA was used to optimize the structure of the FBNN. The organic loading rate, the pH, the microorganisms' concentrations, the hydraulic retention time (HRT), and the total volumetric gas flow rate were the inputs of the ANN model. The proposed model was evaluated in terms of its estimation of the key quality parameters of the reactor, such as the hydrogen production rate and the metabolites in the effluent. The simulation results showed that the FBNN model was able to effectively describe the daily variations of the packed-bed bioreactor performance at various HRTs.