Estimation of Water Demand in Residential Building Using Machine Learning Approach

Abstract

This paper shows an estimation model for residential water consumption using machine learning approach in Korea. We verify the diversifying elements constituting apartment buildings as input datasets for the Back- Propagation Neural Network (BPNN), the most novel supervised learning neural network based model in accordance with the empirical water use data. A water use prediction for residential buildings is a complex and nonlinear function of geographic, climatic, and morphological variables of buildings. For the verification purpose, empirical data sets consisting of water usage data retrieved from multiple residential apartment buildings in Korea were analyzed as case studies. The proposed model accurately forecast water uses for each examined residential apartment buildings. The results of the proposed models could offer a reliable water supply to meet the useful needs of customers and the local community while facilitating the efficient consumption of water.