eduzhai > Applied Sciences > Engineering >

Optimal control towards sustainable wastewater treatment plants based on multi-agent reinforcement learning

  • king
  • (0) Download
  • 20210507
  • Save

... pages left unread,continue reading

Document pages: 34 pages

Abstract: Wastewater treatment plants are designed to eliminate pollutants andalleviate environmental pollution. However, the construction and operation ofWWTPs consume resources, emit greenhouse gases (GHGs) and produce residualsludge, thus require further optimization. WWTPs are complex to control andoptimize because of high nonlinearity and variation. This study used a noveltechnique, multi-agent deep reinforcement learning, to simultaneously optimizedissolved oxygen and chemical dosage in a WWTP. The reward function wasspecially designed from life cycle perspective to achieve sustainableoptimization. Five scenarios were considered: baseline, three differenteffluent quality and cost-oriented scenarios. The result shows thatoptimization based on LCA has lower environmental impacts compared to baselinescenario, as cost, energy consumption and greenhouse gas emissions reduce to0.890 CNY m3-ww, 0.530 kWh m3-ww, 2.491 kg CO2-eq m3-ww respectively. Thecost-oriented control strategy exhibits comparable overall performance to theLCA driven strategy since it sacrifices environmental bene ts but has lowercost as 0.873 CNY m3-ww. It is worth mentioning that the retrofitting of WWTPsbased on resources should be implemented with the consideration of impacttransfer. Specifically, LCA SW scenario decreases 10 kg PO4-eq ineutrophication potential compared to the baseline within 10 days, whilesignificantly increases other indicators. The major contributors of eachindicator are identified for future study and improvement. Last, the authordiscussed that novel dynamic control strategies required advanced sensors or alarge amount of data, so the selection of control strategies should alsoconsider economic and ecological conditions.

Please select stars to rate!


0 comments Sign in to leave a comment.

    Data loading, please wait...