Intelligent control is a promising approach to achieve
stable and
sustainable operation at municipal wastewater treatment plants (WWTPs).
A desirable WWTP intelligent control system can be responsive to influent
dynamics and adaptable for complex multi-objective optimization. In
this study, we developed a novel intelligent control framework based
on machine learning methods, which comprises a prediction module and
control module. The stacking ensemble learning model (SELM) and Q-learning
model (QLM) were used to capture influent dynamics and intelligently
identify optimal parameters, respectively. This SELM–QLM framework
was trained and validated with historical monitoring data archived
at a full-scale WWTP to optimize the nitrogen removal process. The
results showed that control parameters were frequently adjusted in
response to influent variation and energy consumption of aeration,
and the sludge returning process was effectively decreased while maintaining
the stability of effluent total nitrogen (TN) (TN decreased by 19.53%
and energy consumption decreased by 10.37%). Specifically, the SELM
provided accurate predictions of TN concentration without increasing
the data set scale, and the QLM showed superior ability in determining
the optimal solution from nearly contradictory objectives. This study
provides a framework with significant application values for improving
WWTP management inspired by the objective of stable and sustainable
operation.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.