Prediction of Urban Domestic Water Consumption Considering Uncertainty

Abstract

Quantitative predictions of urban domestic water consumption are of great significance for the planning and management of water resources. Aimed at the uncertainty in the process of water consumption forecasting, the kernel density estimation-fractional order reverse accumulative gray model was proposed. Based on the residual error of the fractional order reverse accumulative gray model, the kernel density estimation method was employed for the frequency analysis. According to the design value of residual error at different confidence levels and the predicted value of the fractional order reverse accumulative gray model, the prediction interval of the future water consumption was constructed. The model was validated using the annual urban domestic water consumption data in Beijing, Chongqing, and Qingdao. To test the performance of the model, the model was compared with the gray model with respect to the kernel density estimation—fractional order forward accumulative, kernel density estimation—first-order reverse accumulative, upper and lower bounds partition method—fractional order reverse accumulative and linear model with interval autoregression model. The technique for order of preference by similarity to the ideal solution (TOPSIS) was used to select the optimal model. The results show that the proposed model demonstrates the best performance regarding urban domestic water consumption prediction and can provide powerful decision support for addressing regional urban water consumption forecast issues in the water source sector. The proposed model also provides a new method for the prediction of urban domestic water consumption and other water consumption predictions.