Prediction of the sludge consumption rate of worm reactor and optimization of its operating conditions by applying ANFIS

Aquatic worms' predation is an effective process for ecological sludge reduction, but the sludge reduction efficiency can be affected by the fluctuation of operational conditions. In this study, dissolved oxygen (DO) concentration, temperature, worm density and sludge load were identified to be the main parameters in sludge consumption process. With these parameters, the adaptive neurofuzzy inference system (ANFIS) and artificial neural network (ANN) were employed to predict the sludge consumption rate in the worm reactor. Compared with the ANN model, the ANFIS model was proved to be more effective with the correlation coefficient of 0.82, the mean absolute percentage error of 71.5% and the root mean square error of 16.7. Based on the ANFIS model, the optimal operation conditions were found to be DO of 1.8~3.1 mg·L^-1, temperature of 18.4~21.7°C, worm density below 1.7 g·cm^-2 and sludge load of 563~734 mg·g^-1. Under these conditions, the sludge consumption rate can be maintained above 100 mg·g^-1·d^-1.