Non-Intrusive Load Monitoring by Voltage–Current Trajectory Enabled Transfer Learning
The DOI matches its Crossref record; 40 of the 40 references it cites were found in a registry or an index.
Abstract
Non-intrusive load monitoring (NILM) is a technique for analyzing changes in the voltage and current flowing through the main feeder and determining the appliances in operation as well as their energy consumption. With the increase in amount and type of electric loads nowadays, it is of increasing significance to extract unique load signatures and build robust classification models for NILM. However, the electric loads of different households differ materially from one another, which makes it difficult to collect enough label data and train classification models with strong representation and generalization ability. In this paper, a voltage-current (V-I) trajectory enabled transfer learning method has been proposed for NILM. Different from the existing methods, a deep learning model pretrained on a visual recognition dataset is transferred to train the classifier for NILM, linking the knowledge between different domains. Moreover, the V-I trajectory is also transferred to visual representation by color encoding, which not only enhances the load signature's uniqueness but also enables the NILM implementation of transfer learning. The experimental results on NILM datasets show that the proposed method significantly improves the accuracy and can be efficiently generalized compared with state-of-the-art methods.
Overview
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Integrity signals
Registry
Cites
Statistics
Verification
Claims the paper makes
The claims and statistics of this paper haven’t been checked yet.
Free: it reads the open-access text and checks the first 5 claims that cite a source we can check. Create a free account to check every claim.
Self-check
Percentages, sums and reported differences re-computed from the paper’s own numbers.
The self-check needs the paper’s full text.
Citations
Cites
Deep Residual Learning for Image Recognition
He, K., Zhang, X., Ren, S., & Sun, J. (2016) · 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2016.90 · Cited by 235,474 · cited 1×
“Although no pretrained model has been proposed for NILM, they are not rare for visual recognition tasks such as AlexNet [21], VGG-16 [22], GoogLeNet [23], and ResNet-50 [24].”
MethodVery Deep Convolutional Networks for Large-Scale Image Recognition
Simonyan, K., & Zisserman, A. (2014) · arXiv · doi:10.48550/arxiv.1409.1556 · Cited by 114,148 · cited 1×
“Although no pretrained model has been proposed for NILM, they are not rare for visual recognition tasks such as AlexNet [21], VGG-16 [22], GoogLeNet [23], and ResNet-50 [24].”
MethodImageNet classification with deep convolutional neural networks
Krizhevsky, A., Sutskever, I., & Hinton, G. E. (2017) · Communications of the ACM · doi:10.1145/3065386 · Cited by 112,155 · cited 4×
“AlexNet [21] is a large, deep convolutional neural network and its network architecture is described in Fig.”
BackgroundMethodInfluentialImageNet: A large-scale hierarchical image database
Deng, J., Dong, W., Socher, R., Li, L.-J., Li, K., & Fei-Fei, L. (2009) · 2009 IEEE Conference on Computer Vision and Pattern Recognition · doi:10.1109/cvpr.2009.5206848 · Cited by 65,333 · cited 6×
“Reference [26] transfers the knowledge from ImageNet [28], an object classification dataset, to predict poverty mapping with the help of satellite images.”
MethodInfluentialGoing deeper with convolutions
Szegedy, C., Liu, W., Jia, Y., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., & Rabinovich, A. (2015) · 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR) · doi:10.1109/cvpr.2015.7298594 · Cited by 48,562
Pan, S. J., & Yang, Q. (2010) · IEEE Transactions on Knowledge and Data Engineering · doi:10.1109/tkde.2009.191 · Cited by 23,387 · cited 1×
“However, transferring knowledge between two unrelated or weakly related domain could lead to bad performance, which is called ‘negative transfer’ [20].”
BackgroundA Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection
Kohavi, R. (1995) · Zenodo · doi:10.5281/zenodo.19712698 · Cited by 10,909
Nonintrusive appliance load monitoring
Hart, G. W. (1992) · Proceedings of the IEEE · doi:10.1109/5.192069 · Cited by 3,165 · cited 2×
“As proposed in [5] and defined in [45], type I loads only have two operation states on and off, which is a suitable model for most household appliances, such as a light bulb, toaster, or water pump.”
BackgroundNon-Intrusive Load Monitoring Approaches for Disaggregated Energy Sensing: A Survey
Zoha, A., Gluhak, A., Imran, M., & Rajasegarar, S. (2012) · Sensors · doi:10.3390/s121216838 · Cited by 954 · cited 1×
“As proposed in [5] and defined in [45], type I loads only have two operation states on and off, which is a suitable model for most household appliances, such as a light bulb, toaster, or water pump.”
BackgroundEstimating classification error rate: Repeated cross-validation, repeated hold-out and bootstrap
Kim, J.-H. (2009) · Computational Statistics & Data Analysis · doi:10.1016/j.csda.2009.04.009 · Cited by 792 · cited 1×
“applied as recommended in [43], which means the procedure of K-fold cross-validation will be repeated for R times using random sampling with replacement.”
MethodTransfer Learning from Deep Features for Remote Sensing and Poverty Mapping
Xie, M., Jean, N., Burke, M., Lobell, D., & Ermon, S. (2016) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v30i1.9906 · Cited by 455
Load Signature Study—Part I: Basic Concept, Structure, and Methodology
Liang, J., Ng, S. K. K., Kendall, G., & Cheng, J. W. M. (2010) · IEEE Transactions on Power Delivery · doi:10.1109/tpwrd.2009.2033799 · Cited by 453 · cited 1×
“form, which makes use of the difference between two consecutive snapshots and meets the feature-additive criterion [37].”
BackgroundExploiting HMM Sparsity to Perform Online Real-Time Nonintrusive Load Monitoring
Makonin, S., Popowich, F., Bajic, I. V., Gill, B., & Bartram, L. (2016) · IEEE Transactions on Smart Grid · doi:10.1109/tsg.2015.2494592 · Cited by 378 · cited 1×
“vised methods such as hidden Markov models and Bayesian inference [16], [17] are superior in data clustering but less accurate in classification.”
MethodA Novel Method to Construct Taxonomy Electrical Appliances Based on Load Signaturesof
Lam, H., Fung, G., & Lee, W. (2007) · IEEE Transactions on Consumer Electronics · doi:10.1109/tce.2007.381742 · Cited by 324 · cited 2×
“Reference [29] introduces shape features based on the V-I trajectories to develop a taxonomy of home appliances.”
BackgroundMethodAn Empirical Investigation of V-I Trajectory Based Load Signatures for Non-Intrusive Load Monitoring
Hassan, T., Javed, F., & Arshad, N. (2014) · IEEE Transactions on Smart Grid · doi:10.1109/tsg.2013.2271282 · Cited by 322 · cited 1×
“Reference [30] briefly reviews the wave-shape features and seven trajectory features are summarized.”
BackgroundNon-Intrusive Load Disaggregation Using Graph Signal Processing
He, K., Stankovic, L., Liao, J., & Stankovic, V. (2018) · IEEE Transactions on Smart Grid · doi:10.1109/tsg.2016.2598872 · Cited by 306 · cited 2×
“These two appliances are best represented by multi-state models, which is still an open issue in this field [4], [19].”
BackgroundToward Non-Intrusive Load Monitoring via Multi-Label Classification
Tabatabaei, S. M., Dick, S., & Xu, W. (2017) · IEEE Transactions on Smart Grid · doi:10.1109/tsg.2016.2584581 · Cited by 304
Appliance classification using VI trajectories and convolutional neural networks
De Baets, L., Ruyssinck, J., Develder, C., Dhaene, T., & Deschrijver, D. (2018) · Energy and Buildings · doi:10.1016/j.enbuild.2017.09.087 · Cited by 239 · cited 6×
“The main reason is that the V-I trajectory can be transferred to visual representation and then used in appliance recognition [6], [10], [29]–[33].”
BackgroundMethodInfluentialDistant Domain Transfer Learning
Tan, B., Zhang, Y., Pan, S. J., & Yang, Q. (2017) · Proceedings of the AAAI Conference on Artificial Intelligence · doi:10.1609/aaai.v31i1.10826 · Cited by 214 · cited 2×
“Still, we can bridge two seemingly unrelated domains by one or more intermediate domains, which is called ‘transitive transfer learning’ and has attracted the attention recently [25]–[27].”
BackgroundMethodHome electrical signal disaggregation for non-intrusive load monitoring (NILM) systems
Figueiredo, M. B., de Almeida, A., & Ribeiro, B. (2012) · Neurocomputing · doi:10.1016/j.neucom.2011.10.037 · Cited by 201
Non-intrusive load monitoring through home energy management systems: A comprehensive review
Hosseini, S. S., Agbossou, K., Kelouwani, S., & Cardenas, A. (2017) · Renewable and Sustainable Energy Reviews · doi:10.1016/j.rser.2017.05.096 · Cited by 200 · cited 1×
“NILM not only detects the appliances in use but also have the potential to infer human behaviors and life patterns such as routine times and private activities [46].”
BackgroundApplication of load monitoring in appliances’ energy management – A review
Abubakar, I., Khalid, S. N., Mustafa, M. W., Shareef, H., & Mustapha, M. (2017) · Renewable and Sustainable Energy Reviews · doi:10.1016/j.rser.2016.09.064 · Cited by 191 · cited 1×
“such as smart meter, smart plug, current sensor, and voltage sensor [36].”
BackgroundComprehensive feature selection for appliance classification in NILM
Sadeghianpourhamami, N., Ruyssinck, J., Deschrijver, D., Dhaene, T., & Develder, C. (2017) · Energy and Buildings · doi:10.1016/j.enbuild.2017.06.042 · Cited by 185 · cited 2×
“In order to allow more detailed analysis than mere accuracy, a specific table layout called confusion matrix [33], [40] is used to visualize the performance of this classification problem.”
MethodTan, B., Song, Y., Zhong, E., & Yang, Q. (2015) · Proceedings of the 21th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining · doi:10.1145/2783258.2783295 · Cited by 184 · cited 2×
“Still, we can bridge two seemingly unrelated domains by one or more intermediate domains, which is called ‘transitive transfer learning’ and has attracted the attention recently [25]–[27].”
BackgroundLin, Y.-H., & Tsai, M.-S. (2015) · IEEE Transactions on Smart Grid · doi:10.1109/tsg.2015.2388492 · Cited by 183 · cited 1×
“In order to improve the generalization ability, the ensemble of multiple learning algorithms is proposed and usually obtains better classification performance than that of any learning algorithm alone [13].”
MethodNon-intrusive load monitoring algorithm based on features of V–I trajectory
Wang, A. L., Chen, B. X., Wang, C. G., & Hua, D. (2018) · Electric Power Systems Research · doi:10.1016/j.epsr.2017.12.012 · Cited by 183 · cited 2×
“The main reason is that the V-I trajectory can be transferred to visual representation and then used in appliance recognition [6], [10], [29]–[33].”
BackgroundDenoising autoencoders for Non-Intrusive Load Monitoring: Improvements and comparative evaluation
Bonfigli, R., Felicetti, A., Principi, E., Fagiani, M., Squartini, S., & Piazza, F. (2018) · Energy and Buildings · doi:10.1016/j.enbuild.2017.11.054 · Cited by 179 · cited 1×
“Besides, neural networks can also be deepened by stacking multiple network layers together to improve the representation ability [14], [15].”
BackgroundDeep Sparse Coding for Non–Intrusive Load Monitoring
Singh, S., & Majumdar, A. (2018) · IEEE Transactions on Smart Grid · doi:10.1109/tsg.2017.2666220 · Cited by 158 · cited 1×
“Besides, neural networks can also be deepened by stacking multiple network layers together to improve the representation ability [14], [15].”
BackgroundA feasibility study of automated plug-load identification from high-frequency measurements
Gao, J., Kara, E. C., Giri, S., & Berges, M. E. (2015) · IEEE Global Conference on Signal and Information Processing · doi:10.1109/globalsip.2015.7418189 · Cited by 139 · cited 3×
“From the aspect of computing cost, the time complexity of binary trajectory mapping and color encoding are both O(J) (J is the number of the trajectory points), which equals to the complexity of feature extraction in [10], [31], and [33].”
BackgroundEvent-Based Energy Disaggregation Algorithm for Activity Monitoring From a Single-Point Sensor
Alcalá, J., Ureña, J., Hernández, Á., & Gualda, D. (2017) · IEEE Transactions on Instrumentation and Measurement · doi:10.1109/tim.2017.2700987 · Cited by 134 · cited 1×
“However, they are both public datasets of high-resolution electrical appliance measurements and have been used for NILM’s evaluation in recent work [33], [40], [41].”
MethodA Cloud-Based On-Line Disaggregation Algorithm for Home Appliance Loads
Mengistu, M. A., Girmay, A. A., Camarda, C., Acquaviva, A., & Patti, E. (2018) · IEEE Transactions on Smart Grid · doi:10.1109/tsg.2018.2826844 · Cited by 120
Singhal, V., Maggu, J., & Majumdar, A. (2018) · IEEE Transactions on Smart Grid · doi:10.1109/tsg.2018.2815763 · Cited by 112 · cited 1×
“In order to improve energy efficiency, it is essential to provide fine-grained energy usage information for smart building applications such as energy management and home automation [2]–[4].”
BackgroundDistant domain transfer learning
Tan, B. (2017) · doi:10.14711/thesis-991012535962503412 · Cited by 110
Non-Intrusive Load Monitoring Using Semi-Supervised Machine Learning and Wavelet Design
Gillis, J. M., & Morsi, W. G. (2016) · IEEE Transactions on Smart Grid · doi:10.1109/tsg.2016.2532885 · Cited by 108
Multi-appliance recognition system with hybrid SVM/GMM classifier in ubiquitous smart home
Lai, Y.-H., Lai, C., Huang, Y., & Chao, H. (2012) · Information Sciences · doi:10.1016/j.ins.2012.10.002 · Cited by 99 · cited 1×
“Supervised methods such as k-nearest neighbor, support vector machine, decision tree, neural network are studied in [9]–[12].”
MethodOn the Bayesian optimization and robustness of event detection methods in NILM
Baets, L. D., Ruyssinck, J., Develder, C., Dhaene, T., & Deschrijver, D. (2017) · Energy and Buildings · doi:10.1016/j.enbuild.2017.03.061 · Cited by 79 · cited 1×
“vised methods such as hidden Markov models and Bayesian inference [16], [17] are superior in data clustering but less accurate in classification.”
MethodLin, S., Zhao, L., Li, F. F., Liu, Q., Li, D., & Fu, Y. (2016) · Electric Power Systems Research · doi:10.1016/j.epsr.2015.12.014 · Cited by 74
A Distinctive Load Feature Extraction Based on Fryze's Time-domain Power Theory
Teshome, D. F., Huang, T. D., & Lian, K. L. (2016) · IEEE Power and Energy Technology Systems Journal · doi:10.1109/jpets.2016.2559507 · Cited by 53 · cited 1×
“In this paper, Fryze power theory is introduced to preprocess the V-I trajectory [38].”
MethodAdmittance-based load signature construction for non-intrusive appliance load monitoring
Liu, Y., Wang, X., Zhao, L., & Liu, Y. (2018) · Energy and Buildings · doi:10.1016/j.enbuild.2018.04.049 · Cited by 50 · cited 1×
“Moreover, the consumers also have the choice of energy consumption behaviors according to the suggestions pushed by providers [8].”
BackgroundSelf-citationDesigning New Orthogonal High-Order Wavelets for Nonintrusive Load Monitoring
Gillis, J. M., Chung, J., & Morsi, W. G. (2017) · IEEE Transactions on Industrial Electronics · doi:10.1109/tie.2017.2739701 · Cited by 25 · cited 1×
“Non-intrusive load monitoring (NILM), a concept developed originally by Hart [5], has attracted the attention of researchers recently [6], [7].”
Background
Cited by
Pinto, G., Wang, Z., Roy, A., Hong, T., & Capozzoli, A. (2022) · Advances in Applied Energy · doi:10.1016/j.adapen.2022.100084 · Cited by 295
Liaqat, R., Sajjad, I. A., Waseem, M., & Alhelou, H. H. (2021) · IEEE Access · doi:10.1109/access.2021.3123196 · Cited by 39
Metadata
- Authors
- Liu, Y., Wang, X., & You, W.
- Venue
- IEEE Transactions on Smart Grid
- Published
- 2019-09
- Volume
- 10
- Issue
- 5
- Pages
- 5609-5619
- ISSN
- 1949-3053 (ISSN-L)
- 1949-3061
- OpenAlex
- W2904841534
- Cited by
- 248
- References
- 40
- FWCI
- 5.46
- Type
- journal article
- Language
- en
- Open access
- closed
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
Data sources
Bibliographic data: Semantic Scholar, OpenAlex, Crossref.
Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).
Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.