Energy and Materials-Saving Management via Deep Learning for Wastewater Treatment Plants

Abstract

With the increasing public attention on sustainability, conservation of energy and materials has been a general demand for wastewater treatment plants (WWTPs). To meet the demand, efficient optimal management and decision mechanism are expected to reasonably configure resource of energy and materials.In recent years, advanced computational techniques such as neural networks and genetic algorithm provided data-driven solutions to overcome some industrial problems. They work from the perspective of statistical learning, mining invisible latent rules from massive data. This paper proposes energy and materials-saving management via deep learning for WWTPs, using real-world business data of a wastewater treatment plant located in Chongqing, China. Treatment processes are modeled through neural networks, and materials cost that satisfies single indexes can be estimated on this basis. Then, genetic algorithm is selected as the decision scheme to compute overall cost that is able to simultaneously satisfy all the indexes. Empirically, experimental results evaluate that with the proposed management method, total energy and materials cost can be reduced by 10%–15%.