A PSO-RBF neural network for BOD multi-step prediction in wastewater treatment process

Abstract

The single-step prediction of the effluent biochemical oxygen demand (BOD) in wastewater treatment process (WWTP) cannot provide the information for future trends, which will affect the decision-making for subsequent control strategies in WWTP. To solve this problem, a radial basis function (RBF) neural network soft-sensing model based on PSO algorithm (PSO-RBF) is proposed for multi-step prediction of the biochemical oxygen demand (BOD) in WWTP. First, we propose an RBF model for multi-step prediction based on MIMO strategy. Then, the PSO algorithm is used to optimize the RBF parameters. Finally, the proposed model is applied to a four-step BOD prediction in WWTP. Experimental results show that this model can better predict BOD in multiple steps compared to traditional models.