Modeling of SBR aerobic granular sludge using neural network with GSA and IW-PSO

This paper presents a modeling technique of sequential batch reactor (SBR) for aerobic granular sludge (AGS) using artificial neural network (ANN). A SBR fed with synthetic wastewater was operated at high temperature of 50˚C to study the formation of AGS for simultaneous organics and nutrients removal in 60 days. The feed forward neural network (FFNN) was used to model the nutrients removal process. In this work, inertia weight particle swarm optimization (PSO) and gravitational search algorithm (GSA) were employed to optimize the neural network weights and biases. It was observed that the inertia weight GSA-NN give better prediction of nutrient removal compared with Inertia weight PSO. The performance of the models was measured using the R², mean square error (MSE) and root mean square error (RMSE).