Fuzzy and Neuro-Fuzzy Modelling for Prediction of Effluent COD for a Real Scale UASB Reactor Treating Distillery Wastewater

摘要

The high rate anaerobic processes are used for treatment of industrial wastewater. These processes are subjected to many disturbances due to variation in quantity and quality of wastewater. Neuro-fuzzy (ANFIS) and fuzzy models were developed for prediction of effluent COD for an UASB reactor treating distillery wastewater. Different combinations of input parameters were assessed to obtain the ANFIS model. A fuzzy rule based model was developed with same combination of input parameters as ANFIS model. Expert's knowledge, plant operator's opinion and trend of the data were considered during model development. Parameters required for daily monitoring and operation of the plant were used for model development. Statistical parameters like the correlation coefficient (R) and root mean square error (RMSE) were used to assess the performance of these models. The value of 0.8922 and 1334.69 for R and RMSE were observed with the ANFIS model whereas 0.8197 and 4208.34 were obtained with the fuzzy model respectively. These values indicate good agreement between the predicted and observed values of effluent COD and satisfactory performance of the models. The modelling approach discussed in this paper will be useful for assessment of performance of the real scale UASB reactor and to control the operation of the plant.