Résumé
During the last few decades, citizens around western countries became more and more
sensible to energy saving. However, while home electricity consumption is a source of
concern, the means to reduce this consumption are not easy to _nd and implement for
private individuals. Di_erent studies show that displaying the home consumption could
lead to a reduction of electricity use. However, the global consumption doesn't provide
the consumer with su_cient information about what counts for the main part of their
electricity invoice. Systems able to display the energy used by the main appliances would
greatly help the consumer to _nd out which equipment should be replaced and/or which
behavior should be modi_ed. Implementing such systems requires the disaggregation of
the consumption of the main appliances. One solution consists in measuring the global
electricity consumption and extracting the most important information from the general
load curve using signal processing methods and detection algorithms.
The HES-SO Valais-Wallis (University of Applied Sciences Western Switzerland) and the
CSEM (Swiss Center for Electronics and Microtechnology) are currently working in this
direction. To develop recognition algorithms from an aggregated load curve, an acquisition
system able to measure the three phases of a standard household has been built. This
system has been deployed in seven households and has been acquiring data sampled at
1Hz for over two years for the _rst deployment site. In two households Ecowizz [1] plugs
are used to acquire disaggregated data of the main appliances in parallel to the central
measure. In parallel the HES-SO Valais-Wallis is deploying the system in _fty households
for one month.
The collected data allow a better understanding of the main contributors to the load curve
as well as the useful characteristics to recognize them. To _rst tackle the complexity
of the aggregated load curve, a simulator of the main contributors (washing machine,
dishwasher, tumble dryer, oven, stove, etc.) was also created, thus allowing to initially
test the disaggregation algorithms with an a priori knowledge of the contributors. Both
the database of real signals and the simulator are new tools that will allow for new research
and development of algorithms for the analysis of aggregated load curves.