Scattering Transform for Classification in Non-Intrusive Load Monitoring
Nonintrusive Load Monitoring (NILM) uses computational methods to disaggregate and classify electrical appliances signals. The classification is usually based on the power signatures of the appliances obtained by a feature extractor. State-of-the-art results were obtained extracting NILM features wi...
| Main Authors: | , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2021-10-01
|
| Series: | Energies |
| Subjects: | |
| Online Access: | https://www.mdpi.com/1996-1073/14/20/6796 |
| _version_ | 1797514635477254144 |
|---|---|
| author | Everton Luiz de Aguiar André Eugenio Lazzaretti Bruna Machado Mulinari Daniel Rodrigues Pipa |
| author_facet | Everton Luiz de Aguiar André Eugenio Lazzaretti Bruna Machado Mulinari Daniel Rodrigues Pipa |
| author_sort | Everton Luiz de Aguiar |
| collection | DOAJ |
| description | Nonintrusive Load Monitoring (NILM) uses computational methods to disaggregate and classify electrical appliances signals. The classification is usually based on the power signatures of the appliances obtained by a feature extractor. State-of-the-art results were obtained extracting NILM features with convolutional neural networks (CNN). However, it depends on the training process with large datasets or data augmentation strategies. In this paper, we propose a feature extraction strategy for NILM using the <i>Scattering Transform</i> (ST). The ST is a convolutional network analogous to CNN. Nevertheless, it does not need a training process in the feature extraction stage, and the filter coefficients are analytically determined (not empirically, like CNN). We perform tests with the proposed method on different publicly available datasets and compare the results with state-of-the-art deep learning-based and traditional approaches (including wavelet transform and V-I representations). The results show that ST classification accuracy is more robust in terms of waveform parameters, such as signal length, sampling frequency, and event location. Besides, ST overcame the state-of-the-art techniques for single and aggregated loads (accuracies above 99% for all evaluated datasets), in different training scenarios with single and aggregated loads, indicating its feasibility in practical NILM scenarios. |
| first_indexed | 2024-03-10T06:34:23Z |
| format | Article |
| id | doaj.art-22b96e712f074e9ea2f673fff73d7a1a |
| institution | Directory Open Access Journal |
| issn | 1996-1073 |
| language | English |
| last_indexed | 2024-03-10T06:34:23Z |
| publishDate | 2021-10-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Energies |
| spelling | doaj.art-22b96e712f074e9ea2f673fff73d7a1a2023-11-22T18:08:55ZengMDPI AGEnergies1996-10732021-10-011420679610.3390/en14206796Scattering Transform for Classification in Non-Intrusive Load MonitoringEverton Luiz de Aguiar0André Eugenio Lazzaretti1Bruna Machado Mulinari2Daniel Rodrigues Pipa3CPGEI—Graduate Program in Electrical and Computer Engineering, Universidade Tecnológica Federal do Paraná—UTFPR, Sete de Setembro, 3165, Curitiba 80230-901, PR, BrazilCPGEI—Graduate Program in Electrical and Computer Engineering, Universidade Tecnológica Federal do Paraná—UTFPR, Sete de Setembro, 3165, Curitiba 80230-901, PR, BrazilDataplai, Eng. Niepce da Silva, 200, Curitiba 80610-280, PR, BrazilCPGEI—Graduate Program in Electrical and Computer Engineering, Universidade Tecnológica Federal do Paraná—UTFPR, Sete de Setembro, 3165, Curitiba 80230-901, PR, BrazilNonintrusive Load Monitoring (NILM) uses computational methods to disaggregate and classify electrical appliances signals. The classification is usually based on the power signatures of the appliances obtained by a feature extractor. State-of-the-art results were obtained extracting NILM features with convolutional neural networks (CNN). However, it depends on the training process with large datasets or data augmentation strategies. In this paper, we propose a feature extraction strategy for NILM using the <i>Scattering Transform</i> (ST). The ST is a convolutional network analogous to CNN. Nevertheless, it does not need a training process in the feature extraction stage, and the filter coefficients are analytically determined (not empirically, like CNN). We perform tests with the proposed method on different publicly available datasets and compare the results with state-of-the-art deep learning-based and traditional approaches (including wavelet transform and V-I representations). The results show that ST classification accuracy is more robust in terms of waveform parameters, such as signal length, sampling frequency, and event location. Besides, ST overcame the state-of-the-art techniques for single and aggregated loads (accuracies above 99% for all evaluated datasets), in different training scenarios with single and aggregated loads, indicating its feasibility in practical NILM scenarios.https://www.mdpi.com/1996-1073/14/20/6796scattering transformNILM featuresfeatures extractor |
| spellingShingle | Everton Luiz de Aguiar André Eugenio Lazzaretti Bruna Machado Mulinari Daniel Rodrigues Pipa Scattering Transform for Classification in Non-Intrusive Load Monitoring Energies scattering transform NILM features features extractor |
| title | Scattering Transform for Classification in Non-Intrusive Load Monitoring |
| title_full | Scattering Transform for Classification in Non-Intrusive Load Monitoring |
| title_fullStr | Scattering Transform for Classification in Non-Intrusive Load Monitoring |
| title_full_unstemmed | Scattering Transform for Classification in Non-Intrusive Load Monitoring |
| title_short | Scattering Transform for Classification in Non-Intrusive Load Monitoring |
| title_sort | scattering transform for classification in non intrusive load monitoring |
| topic | scattering transform NILM features features extractor |
| url | https://www.mdpi.com/1996-1073/14/20/6796 |
| work_keys_str_mv | AT evertonluizdeaguiar scatteringtransformforclassificationinnonintrusiveloadmonitoring AT andreeugeniolazzaretti scatteringtransformforclassificationinnonintrusiveloadmonitoring AT brunamachadomulinari scatteringtransformforclassificationinnonintrusiveloadmonitoring AT danielrodriguespipa scatteringtransformforclassificationinnonintrusiveloadmonitoring |