The identification and adaptive prediction of urban sewer flows

Abstract

The major limitation in this study of the adaptive prediction of urban sewer flows has been the poor quality of the data. In any future study it can be expected that, while pumping disturbances may not be eliminated completely, better data would be available for analysis. With a view to on-line implementation of the predictor it would, therefore, be important to site the flow-measuring equipment at a carefully chosen location. One-step ahead forecasts of the plant influent flow are obtained from an adaptive predictor which closely approaches the satisfaction of the practical constraints on the system: namely, as little automated instrumentation as possible should be assumed. The salient feature of the black box model for the predictor is its simplicity and compactness when compared with other, largely deterministic, models based on the physical laws of the system behaviour. For it should be remembered that the currently existing technology of wastewater treatment favours the simple rather than the sophisticated.