Hybrid data driven approach based on ANNs-PCA for wastewater treatment plant performance assessment

Abstract

In this paper, a data driven method to assess and predict performance of full scale urban activated sludge wastewater treatment plant (WWTP) is presented. The proposed hybrid approach consists of a combination of artificial neural networks (ANNs) and principal component analysis (PCA). Measurement results of a municipal activated sludge WWTP operation of 1.3 million inhabitant equivalents are presented and discussed. In ANNs PCA design, the ANNs used to calculate a nonlinear and dynamic model of the processes under normal operating conditions. Besides, PCA is used to generate monitoring charts based on all measured parameters. Results highlight that ANNs-PCA monitoring is crucial tool that can be used to optimize and predict process spatiotemporal evaluation. This research results provide a practical strategy for improving operation, management and performance prediction of studied WWTP. This supports Sustainable Development Goal (SDG) 6: Clean Water and Sanitation and worldwide sustainability actions and efforts. • Data driven methods represent a powerful tool for decision making in wastewater treatment • Hybrid data driven approach based on ANNs and PCA is presented • ANNs-PCA is used to evaluate and predict performance of full scale WWTP • Developed ANNs-PCA approach can be applied to optimize operation of WWTP • Optimal control of water-based applications using AI is a key factor to ensure sustainability