Activity recognition in smart homes based on electrical devices identification

C Belley, S Gaboury, B Bouchard… - Proceedings of the 6th …, 2013 - dl.acm.org
C Belley, S Gaboury, B Bouchard, A Bouzouane
Proceedings of the 6th international conference on pervasive technologies …, 2013dl.acm.org
Activity recognition constitutes the key challenge in the development of smart home assistive
systems. In this paper, we propose a new algorithmic method for activity recognition in a
smart home, based on load signatures of appliances. Most recognition approaches rely on
distributed and heterogeneous sensors (ex. RFID), which are intrusive require complex
installation, deployment and maintenance. On the other hand, most applications of
appliance load monitoring (signal analysis) refer to the energy saving and the costs …
Activity recognition constitutes the key challenge in the development of smart home assistive systems. In this paper, we propose a new algorithmic method for activity recognition in a smart home, based on load signatures of appliances. Most recognition approaches rely on distributed and heterogeneous sensors (ex. RFID), which are intrusive require complex installation, deployment and maintenance. On the other hand, most applications of appliance load monitoring (signal analysis) refer to the energy saving and the costs reducing of energy consumption. Consequently, our proposal constitutes an original application and new algorithmic method based on steady-state operations and signatures. The extraction process of load signatures of appliances is carried out in a three-dimensional space through a single power analyzer, which is non-intrusive (NIALM). We have rigorously tested this new approach by conducting an experiment in our smart home prototype by simulating daily scenarios taken from clinical trials previously done with Alzheimer patients. The promising results we obtained are presented and compared to other approaches, showing that, with an exceptionally minimal investment and the exploitation of relatively limited data, our method can efficiently recognize activities of daily living for providing assistive services.
ACM Digital Library
顯示最佳搜尋結果。 查看所有結果