Learning Systems for Electric Consumption of Buildings

Abstract

Individual appliances' electricity consumption is automatically disaggregated from a single custom metering system on the main feed to an occupied residential building. A data acquisition system samples voltage and current at 100 kHz, then calculates real and reactive power, harmonics, and other features at 20Hz. A probabilistic eventdetector using the generalized likelihood ratio (GLR) matches human-labeled events to the time-series of features. Machine-learning classification was most successful with the 1-nearest-neighbor algorithm, correctly identifying 90% of the laboratorygenerated training events and 79% of validation examples. The challenge of obtaining adequate training data for the real-world home leads to the development of the Wire Spy, a wirelessly-networked event detector with an inductive sensor which clamps to the cable of an appliance.

Is disaggregation the holy grail of energy efficiency? The case of electricity

“…They implement several existing NALM approaches, effectively experimenting with different appliance signatures proposed in the literature (real and reactive power, harmonics and transients) combined with off-the-shelf machine learning classification algorithms. In Berges et al (2009) the same group proposes a 1-nearest neighbor algorithm on a Euclidian metric for appliance classification, which they use on real and reactive power change events, as well as harmonics detected from voltage and current data sampled at 100 kHz from 8 different…”

“…The second approach that performed beyond its class utilized a competition strategy among multiple algorithms within the system (Berges et al, 2009. It matched each new unidentified appliance signature to a library value (i.e., a known appliance signature in a database of such signatures) using several different algorithms, and the one that produced the best match "won".…”

The user side of sustainability: Modeling behavior and energy usage in the home

“…Non-intrusive appliance load monitoring [10] has been designed to detect the turning on and off of individual appliances in an electrical circuit. Several academic studies focused on this topic to estimate residential energy levels based on appliance usage [11,12]. With respect to energy conservation, some industrial products focus on providing energy information services and saving tips to residents.…”

A low-complexity energy disaggregation method: Performance and robustness

“…Supervised NALM methods use labeled appliance events to train classifiers, and are usually based on optimization and pattern recognition approaches, such as rule-based, SVM or Bayes-based classification. Unsupervised methods do not require labeled sets and are usually based on clustering [19] or HMMs [8]- [10].…”