Multivariate Non-Negative Matrix Factorization with Application to Energy Disaggregation

Schirmer, Pascal A.; Mporas, Iosif

doi:10.1109/icassp39728.2021.9414202

Cited by 4 publications

(3 citation statements)

References 17 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…For example, in [23], the authors show how different features improve on the results of the identification of the chosen TV channel based only on the consumption of the TV. In [24], the authors extend matrix factorization to many features (active, reactive, apparent, and current) and show that using more features provide better results than the one-feature approach in [25]. Additionally, the book in [18] contains experiments about the evaluation of features when combined with specific approaches for energy disaggregation.…”

Section: Related Workmentioning

confidence: 99%

Evaluation of Features Generated by a High-End Low-Cost Electrical Smart Meter

et al. 2021

View full text Add to dashboard Cite

The problem of energy disaggregation is the separation of an aggregate energy signal into the consumption of individual appliances in a household. This is useful, since the goal of energy efficiency at the household level can be achieved through energy-saving policies towards changing the behavior of the consumers. This requires as a prerequisite to be able to measure the energy consumption at the appliance level. The purpose of this study is to present some initial results towards this goal by making heavy use of the characteristics of a particular din-rail meter, which is provided by Meazon S.A. Our thinking is that meter-specific energy disaggregation solutions may yield better results than general-purpose methods, especially for sophisticated meters. This meter has a 50 Hz sampling rate over 3 different lines and provides a rather rich set of measurements with respect to the extracted features. In this paper we aim at evaluating the set of features generated by the smart meter. To this end, we use well-known supervised machine learning models and test their effectiveness on certain appliances when selecting specific subsets of features. Three algorithms are used for this purpose: the Decision Tree Classifier, the Random Forest Classifier, and the Multilayer Perceptron Classifier. Our experimental study shows that by using a specific set of features one can enhance the classification performance of these algorithms.

show abstract

Section: Related Workmentioning

confidence: 99%

Evaluation of Features Generated by a High-End Low-Cost Electrical Smart Meter

et al. 2021

View full text Add to dashboard Cite

show abstract

“…First, pattern matching (elastic matching) techniques, which are detecting device signatures in the aggregated power consumption signal have been proposed [2]- [5]. Second, source separation methods, such as matrix and tensor factorization as well as sparse coding, have been utilized separating base components and activations [6]- [9]. Third, machine learning and deep learning based models have been used to generate data driven models to estimate the power consumption of devices from the aggregated signal [10]- [13].…”

Section: Introductionmentioning

confidence: 99%

Device and Time Invariant Features for Transferable Non-Intrusive Load Monitoring

Schirmer

Mporas

2022

IEEE Open J. Power Energy

Self Cite

View full text Add to dashboard Cite

Non-Intrusive Load Monitoring aims to extract the energy consumption of individual electrical appliances through disaggregation of the total power consumption as measured by a single smart meter in a household. Although when data from the same household are used to train a disaggregation model the device disaggregation accuracy is quite high (80% -95%), depending on the number of devices, the use of pre-trained disaggregation models in new households in most cases results in a significant reduction of disaggregation accuracy. In this article we propose a transferability approach for Non-Intrusive Load Monitoring using fractional calculus and normalized Karhunen Loeve Expansion based spectrograms followed by a Convolutional Neural Network in order to generate device characteristic features that do not change significantly across different households. The performance of the proposed methodology was evaluated using two publicly available datasets, namely REDD and REFIT. The proposed transferability approach improves the Mean Absolute Error by 13.1% when compared to other transfer learning approaches for energy disaggregation.

show abstract

“…The aim of Non-Intrusive Load Monitoring (NILM) is to estimate the power consumption on the device level from the aggregated power-consumption signal of a household or a building [ 1 ], while minimizing the number of installed energy meters and thus reducing the wiring harness and improving the retrofitting capabilities [ 1 , 2 ]. NILM is defined as a single-channel source-separation task, and the methods that have been proposed in the literature to solve it can be classified into three main categories [ 3 ], namely (i) the pattern-matching (elastic matching) approaches which detect load signatures in the aggregated power-consumption signal by comparing them to a set of reference signatures [ 4 , 5 , 6 ]; (ii) the source-separation methods, including matrix and tensor factorization as well as sparse coding, which separate base components and activations using numeric solvers [ 7 , 8 , 9 ]; and (iii) the model-based approaches which are based on machine learning algorithms, usually training one model per device, in order to estimate the power consumption of the loads of interest from the aggregated signal [ 10 , 11 , 12 ].…”

Section: Introductionmentioning

confidence: 99%

2D Transformations of Energy Signals for Energy Disaggregation

Schirmer

Mporas

2022

Sensors

Self Cite

View full text Add to dashboard Cite

The aim of Non-Intrusive Load Monitoring is to estimate the energy consumption of individual electrical appliances by disaggregating the overall power consumption that has been sampled from a smart meter at a house or commercial/industrial building. Last decade’s developments in deep learning and the utilization of Convolutional Neural Networks have improved disaggregation accuracy significantly, especially when utilizing two-dimensional signal representations. However, converting time series’ to two-dimensional representations is still an open challenge, and it is not clear how it influences the performance of the energy disaggregation. Therefore, in this article, six different two-dimensional representation techniques are compared in terms of performance, runtime, influence on sampling frequency, and robustness towards Gaussian white noise. The evaluation results show an advantage of two-dimensional imaging techniques over univariate and multivariate features. In detail, the evaluation results show that: first, the active and reactive power-based signatures double Fourier based signatures, as well as outperforming most of the other approaches for low levels of noise. Second, while current and voltage signatures are outperformed at low levels of noise, they perform best under high noise conditions and show the smallest decrease in performance with increasing noise levels. Third, the effect of the sampling frequency on the energy disaggregation performance for time series imaging is most prominent up to 1.2 kHz, while, above 1.2 kHz, no significant improvements in terms of performance could be observed.

show abstract

Multivariate Non-Negative Matrix Factorization with Application to Energy Disaggregation

Cited by 4 publications

References 17 publications

Evaluation of Features Generated by a High-End Low-Cost Electrical Smart Meter

Evaluation of Features Generated by a High-End Low-Cost Electrical Smart Meter

Device and Time Invariant Features for Transferable Non-Intrusive Load Monitoring

2D Transformations of Energy Signals for Energy Disaggregation

Contact Info

Product

Resources

About