Bio-communal wastewater treatment plant real-time modeling using an intelligent meta-heuristic approach: A sustainable and green ecosystem

Considering the importance of nitrogen and organic carbon in supporting the growth of various algae and organic matters, that improves eutrophication along the water bodies. It is, therefore, essential to develop reliable tools that can help policymakers and experts in-understand the environmental and aquatic importance of these macronutrients in relation to other different hydro-physical and chemical variables. The current research proposes the development of three different data intelligence; Neuro-fuzzy (NF), Support Vector Regression (SVR), and Multivariate Regression (MVR) model integrated with a hybrid Neuro-fuzzy-Shuffled frog-leaping algorithm (NF-SFLA) for the simulation of total Kjeldahl Nitrogen (TKN_eff) and total organic carbon (TCOD_eff) based-organic removal from wastewater treatment plant (WWTP). Both visualized and numerical performance of the single models indicated that NF model combination two (NF-M2) showed superior performance than other models for modeling TKNeff (MAE = 0.0367, NSE = 0.9110) and TCOD_eff (MAE = 0.0363, NSE = 0.8300). The hybrid NF-SFLA was proposed to improve the performance prediction of both TKN_eff and TCOD_eff from the WWTP effluent. The NF-SFLA-M2 achieves an overwhelming performance improvement of up to 13 % over NF-M2 for TKN_eff and TCOD_eff regarding NSE criteria.