Adaptive neuro-fuzzy inference-based modeling of a full-scale expanded granular sludge bed reactor treating corn processing wastewater

Abstract

In this study, an adaptive neuro-fuzzy inference system (ANFIS) including five process variables, such as influent chemical oxygen demand, influent flow rate, influent total Kjeldahl nitrogen, effluent volatile fatty acids and effluent bicarbonate, was described to predict the effluent chemical oxygen demand load from a full-scale expanded granular sludge bed reactor (EGSBR) treating corn processing wastewater. The proposed ANFIS model was conducted by applying hybrid learning algorithm and the model performance was tested by the means of distinct test data set randomly selected from the experimental domain. The ANFIS-based predictions were also validated using various descriptive statistical indicators, such as root mean-square error, index of agreement, the factor of two, fractional variance, proportion of systematic error, etc. The lowest root mean square error (RMSE = 0.03655) and the highest determination coefficient (R 2 = 0.958) were achieved with the subtractive clustering of a first-order Sugeno type fuzzy inference system. ANFIS predicted results were compared with the outputs of multiple nonlinear regression analysis-based models derived in the scope of the present work. Statistical performance indices computed for the testing data set proved that the developed ANFIS-based model exhibited a very good precision in predicting the effluent chemical oxygen demand, load for the EGSBR system. Due to high capability of the ANFIS model in capturing the dynamic behavior and non-linear interactions, it was demonstrated that a complex system, such as anaerobic digestion, could be easily modeled.

Approaches for Modeling Anaerobic Granule-Based Reactors

“…Furthermore, there are 86 parameters in the ADM1. While nonsensitive parameter values can be adopted from the literature, sensitive parameters-which vary significantly-must be calibrated, which is extremely time-consuming and laborious [27]. In addition, the mass of microbial species in bioreactors are not measurable, which challenges the implementation of ADM1 [30].…”

“…Calibrating ANN models is easier than ADM1. When the measured variables begin to show differences in the response of ANN models, the model can be retrained using the newer data employed for cross-checking [27]. Numerous applications of ANNs have been successfully utilized in wastewater treatment modeling [38][39][40].…”

“…Although a anaerobic digestion model can be applied, model calibration is difficult and laborious, while errors between model results and measured results still cannot be ignored [26,27]. The nonlinear regression method shows an empirical relationship between effluent chemical oxygen demand (COD) and operation parameters such as organic load and hydraulic retention time (HRT) [28].…”

Poultry Slaughterhouse Wastewater Treatment Plant Design Advancements

“…The design of the static granular bed reactor (SGBR) is based on the UASB reactor but with a down-flow configuration intended to facilitate the separation of wastewater, solids, and biogas [27]. The expanded granular sludge bed (EGSB) is also founded on the design of the UASB but with better organic loading rates (OLRs), production of gas, soluble pollutants removal, and enhanced mixing in the reactor [28]. These technologies have been widely applied in wastewater treatment [27].…”

“…Developing a mathematical model can contribute to enhancing and monitoring a specific environmental process more efficiently since it allows the role and effect of significant parameters to be examined. The accuracy and dependability of accessible experimental data, the nature of the wastewater being treated together with the biochemical reactions involved determine the value of modelling [28], [37]. The efficacy of anaerobic reactors is affected by reactor flow patterns, loading rates, the existence of toxic compounds, mass transfer in the biofilm and kinetic effects.…”