Forecasting Daily Urban Water Consumption using Conjunctive Evolutionary Algorithm and Wavelet Transform Analysis, A Case Study of Hamedan City, Iran

Water demand forecasting is an important tool in the design, operation, and management of urban water supply systems. The wide variety of factors affecting urban water demand and the variations in the impact levels of these factors due to changes in environmental conditions have undermined the efficiency of conventional mathematical forecasting models in forecasting water demand. Different methods have been so far employed for urban water demand forecasting, among which evolutionary algorithms are the most widely used. In this study, the gene expression programming model, which has a high convergence speed with high precision in calculation and simulation, is combined with the wavelet transform analysis to derive a hybrid model for forecasting daily water demand (consumption) in the city of Hamedan. Water consumption of previous days and climatic parameters constitute the factors affecting water demand in this model. In the first part of the present study, the efficiency of gene expression programming models in forecasting urban daily water demand is investigated to identify the best model (i.e., the best combination of inputs). The second part is dedicated to the evaluation of the effect of wavelet analysis on the results obtained. The results indicate that the best model for forecasting daily water demand is the one with water consumptions of 1, 2, and 3 previous days as well as those of the preceding week as its input. It is also found that the combined gene expression programming and wavelet transform analysis leads to a 10% improvement in forecasting results.