Electricity usage profile disaggregation of hourly smart meter data

Zhao, Bochao and Stankovic, Lina and Stankovic, Vladimir (2018) Electricity usage profile disaggregation of hourly smart meter data. In: 4th International Workshop on Non-Intrusive Load Monitoring, 2018-03-07 - 2018-03-08.

[thumbnail of Zhao-etal-NILM-2018-Electricity-usage-profile-disaggregation-of-hourly-smart-meter]
Text. Filename: Zhao_etal_NILM_2018_Electricity_usage_profile_disaggregation_of_hourly_smart_meter.pdf
Accepted Author Manuscript

Download (409kB)| Preview


This paper is motivated by the growing demand of disaggregating electricity consumption measured by smart meters, down to appliance level. The very low 15-min to 60- min granularity of energy measurements available for analysis, as is standard by the majority of nationwide smart metering programmes, is posing serious challenges. The non-intrusive load monitoring (NILM) solutions for these very low data rates cannot leverage on low (1-60sec) to high rates (in the order of kHz to MHz) NILM approaches, and so far have not received much attention in the literature. In this paper, we propose a novel electricity profile hourly disaggregation of energy consumed (kWh) based on K-nearest neighbours (K-NN), that relies on features such as statistical measures of the energy signal, time usage profile of appliances and reactive power consumption (if available). We propose relative standard deviation as a metric to assess the quality of each feature per appliance. For validation, three publicly accessible real-world datasets are used, namely the REDD, REFIT and AMPds (Version 2), for up to 3 months.