Non-intrusive load monitoring using multi-label classification methods
- PDF / 439,823 Bytes
- 13 Pages / 595.276 x 790.866 pts Page_size
- 24 Downloads / 182 Views
ORIGINAL PAPER
Non-intrusive load monitoring using multi-label classification methods Ding Li1 · Scott Dick1 Received: 10 October 2019 / Accepted: 24 July 2020 © Springer-Verlag GmbH Germany, part of Springer Nature 2020
Abstract Non-intrusive load monitoring is a technique to help power companies monitor and analyze residential energy usage. Aggregated power load measurements for a household (i.e., the signal on the main powerline) are disaggregated into individual appliance loads by examining the appliance-specific power consumption characteristics. These data can then be used to modify consumer behaviors via detailed billing and/or demand-pricing tariffs. A number of advances in the field have been reported in the past two decades, many of which apply machine learning algorithms. However, these algorithms usually only assign one label to an example, which is a poor match to the monitoring problem, meaning elaborate encodings or classifier ensembles are needed. A more elegant solution would be to use algorithms that assign multiple labels to a single example. These multi-label classification algorithms have received very little attention in this field to date. We conduct an experimental investigation of four multi-label classification algorithms for non-intrusive monitoring and find that the best one is superior to the existing reported results on multiple real-world household datasets. Keywords Non-intrusive load monitoring · Load disaggregation · Multi-label classification
List of symbols L The number of household appliances The aggregated power load at time instant t χt N (x) The set of k-nearest neighbors of x − → A label vector for x yx − → The label counting vector of x Cx Hl1
The event that x belongs to label l
Hl0 j El
The event that x not having label l
Ul σd
The set of training instances with the label l The smoothing parameter of Gaussian kernel A scaling factor A (L + 1)-dimensional vector The observable time measurement The reconstructed vector
μ ω R n
B
The event that there are j instances among x’s neighbors assigned label l
Scott Dick [email protected] Ding Li [email protected]
1
Department of Electrical and Computer Engineering, University of Alberta, Edmonton, AB T6G 1H9, Canada
hloss(h) rloss( f ) Yi Yi one_error( f ) γ coverage( f ) avgprec( f ) tp tn fp fn
Hamming loss The symmetric difference between two sets Ranking Loss The true label set for instance xi The complementary set of Yi One error The complete label set for the training dataset Coverage Average precision The number of true positives The number of true negatives The number of false positives The number of false negatives
1 Introduction Electricity demand from residential users usually varies significantly with the time of day and human activities. As a consequence, utilities must devote extensive resources to providing reserve power generation to handle the peak demand in a day. Literally billions of dollars worth of boilers, turbines and generators thus sit idle—or worse, are operating and producing greenhouse
Data Loading...
