Identification of domestic water consumption in a house based on fuzzy clustering algorithms

This work presents the classification of different types of consumptions of water in a house (sinks, showers, washing machines etc.). This classification takes into account the measured flow and the duration of the flow at a particular point in the water distribution network. The classifier uses the FCM and Gustafson-Kessel algorithms. The data set is called AGUA and it corresponds to real data gathered for a research project in Guadalajara, Mexico. The classifier was trained in an unsupervised way. As such, it learns the patterns for the flow and duration of flow, for each type of consumption. The identified classes are relevant consumption types such as using the sink, using the shower, etc. The results show that the proposed approach gives good results, with 91.6 % of the examples classified correctly, and it could be used in the future as part of a supervisory system in order to make better use of water in households.