Simplifying water consumption flow traces for improving end use recognition: a case study

The success of automatic water end use disaggregation and classification strategies greatly depends on the filtering and signal conditioning of the flow traces recorded. The work presented proposes a new filtering algorithm of water consumption flow traces. To improve the performance of the filter, the parameters driving the process are found per event by an automatically calibration procedure. These parameters are selected to ensure the maximum adaptability and simplification of the filtered flow traces. The methodology has been tested with 5210 consumption events obtained from a measurement campaign conducted in a Spanish city. The results obtained show that the filtering algorithm is capable of significantly simplifying the original flow traces while maintaining their main characteristics. On average, it has been found that the most complex events can be described using only 10% of the input data. This analysis can be used to make more efficient the filtering procedure proposed