Parameter Estimation of Seasonal Arima Models for Water Demand Forecasting Using the Harmony Search Algorithm

Accurate short-term water demand forecasts are imperative for the decision making process in water management. It can be applied in the design, future studies, and for possible quality distribution problems. A double seasonal ARIMA model is used to generate water demands forecast (one-day) for a district metering area (DMA). Harmony Search is applied to the parameter estimation of the ARIMA model based on historical water demand data. The Harmony Search optimization algorithm is based on a musical theory process to search for a perfect harmony. A least-squares objective function is formulated by minimizing the random shock series. Real residential water meter readings are used to validate the proposed method by comparing the observed water demand series with the ARIMA model generated time series. The simulation results show a good correspondence to measured water demands primarily on typical days and demonstrate that the harmony search is effective.