Towards hydro-informatics modernization with real-time water consumption classification

Abstract

According to the European Environment agency, water demand across Europe has steadily increased over the past 50 years, partly due to population growth and people moving to cities and towns, especially in densely populated areas. Household use is reported to account for 12% of total water use in Europe. Effective water management practices are being put in place EU-wide, but those that target residential end users are limited to public awareness campaigns, promoting the purchase and use of water-saving devices. Our system aims to bridge the gap between consumers and appliances by accurate and on-time monitoring of household water consumption, at individual appliance level, helping users ease into enduring water-saving practices. Our system’s platform incorporates advanced signal processing methodologies combined with supervised machine learning classifiers to classify water flows, thus identifying residential water appliances with high accuracy. Our experimentation confirms that our models achieve accuracy of ~91% in classifying the four most used household water appliances. This is crucial in assisting end users in reducing their households’ overall water consumption.