Sequential method of change detection and adaptive prediction of municipal water demand

Abstract

An on-line recursive prediction algorithm for peak hourly flow rates of municipal water demands is presented. A simple state space model, which can be identified by using only the past hourly water flow rates time series, is proposed. Prediction accuracies are enhanced by means of sequential changes, detection/acceptance decision rules and an adaptive filter, which revises the gain of the prediction algorithm to respond to the most recent process changes. A simulation test of the algorithm, using real historical water demand data from the city of Columbus, demonstrates the validity of the proposed algorithm.