Publikacje

Abstract:

The publication presents the possibility of modeling in a 1 d advance of the content of organic compounds in the influent wastewater to the treatment plant, where the content of these compounds is determined by both the biochemical and chemical oxygen demand. To predict the quality of the wastewater at the inflow a set of indicators where used to make measurements on a daily basis. In order to develop statistical models 3 methods where used, namely: multivariate adaptive regression splines (MARS), boosted trees (BT), and genetic programming (GP). The carried-out calculations showed that, to calculate the BOD5 there can only be used models developed on the basis
of the value of daily wastewater flow rate to the wastewater treatment plant with 1- and 2-d lags compared to the predicted value. However, in the forecast model of the COD, better wastewater quality index was obtained when as the explanatory variables were measured with COD values of 1- and 3-d lags to the modeled quantity than the daily flow rate referred to the last two measurements.

POWRÓT

Strona główna > Publikacje

Prediction of BOD5 content of the inflow to the treatment plant using different methods of black box – the case study