JAUE2018-082 Modeling the Removal Performance in an ABBR by using Modified BPNN
DOI:
https://doi.org/10.69457/aiue.20180082Keywords:
high-strength organic wastewater, anaerobic baffled biological reactor (ABBR), multiple linear regression (MLR)Abstract
The development and implementation of Modified Back Propagation Neural Network (MBPNN) and Multiple Linear Regression (MLR) models were investigated and compared to evaluate the performance of an anaerobic baffled biological reactor (ABBR) for treating high-strength organic wastewater. Several input variables (COD, BOD, SS, VFA, alkalinity, MLSS, pH, Temperature, and HRT) were measured over a period of 252 days.
The simulation of COD using the MLR and MBPNN models produced R² values of 0.493 and 0.950, respectively, while the simulation of effluent BOD resulted in R² values of 0.583 for MLR and 0.942 for MBPNN. The predicted values of the effluent COD and BOD were in good agreement with the experimental data, with the MBPNN model performing relatively better than the MLR model.It can be readily concluded that, although the MBPNN model is still a relatively young field, it has shown significant potential to improve ABBR operation and predict effluent quantities effectively.
