Gray and Neural Network Prediction of Effluent from the Wastewater Treatment Plant of Industrial Park Using Influent Quality

Abstract

Five types of gray models (GMs) were employed to predict suspended solids (SSeff), chemical oxygen demand (CODeff), and pHeff in the effluent from a wastewater treatment plant (WWTP) in industrial park of Taiwan. For comparison, an artificial neural network (ANN) was also used. Results indicated that the minimum MAPEs of 18.91, 6.10, and 0.86% for SSeff, CODeff, and pHeff could be achieved using GMs. A good fitness could be achieved using ANN also, but they required a large quantity of data for constructing model. Contrarily, GM only required a small amount of data (at least four data), and the prediction results were even better than those of ANN. In the first type of application, the MAPE values for predicting SSeff and pHeff were lower when using GM1N2-1. MAPE value of CODeff using GM1N3-1 was lower when predicting. In the second type, the MAPE value of SSeff using GM (1, 1) was lower when predicting. When predicting CODeff and pHeff, the values using rolling GM (1, 1) (RGM, i.e., four data before the predicted point were used to construct model) were lower. According to the results, the influent indices could be applied on the prediction of effluent quality. It also revealed that GM could predict the industrial effluent variation as its effluent data was insufficient.