Time series distance-based methods for non-intrusive load monitoring in residential buildings

Non-intrusive load monitoring (NILM) deals with the disaggregation of individual appliances from the total load at the smart meter level. This work proposes a generic methodology using temporal sequence classification algorithms. It is based on a low sampling rate unlike other approaches in this domain. An innovative time series distance-based approach in the temporal classification domain is compared with a standard NILM application based on the hidden Markov model (HMM) algorithm. The method is validated over a data-set of 100 houses for a duration of 1 year (with a 10 min sampling rate). A qualitative analysis of the database is also conducted, allowing to segment it into four major clusters based on discussed features.