If you find this repo interesting, I strongly suggest that you check out https://github.com/ihomelab/dnn4nilm_overview and the accompaigning publication Review on Deep Neural Networks Applied to Low-Frequency NILM that compile a much more extensive list of publications. This repo will not be updated.
In order to spare anybody interested in low frequency NILM (i.e. sampling >= 1sec) the arduous task of finding related work, I compiled the following list. Please feel free to pull requests or open an issue.
| Reference | Publication Year | Employed Dataset(s) | Input: Features | Data Augment? | Output: Type [on/off, P] | DNN: Type (Trainable params) | Code |
|---|---|---|---|---|---|---|---|
| Jia Y, Batra N, Wang H, Whitehouse K (2019) A Tree-Structured Neural Network Model for Household Energy Breakdown | 2019 | dataport | P | No | P | dAE, RNN, TreeRNN, CNN, JointCNN, TreeCNN | https://github.com/yilingjia/TreeCNN-for-Energy-Breakdown.git |
| Harell A, Makonin S, Bajić IV (2019) Wavenilm: A causal neural network for power disaggregation from the complex power signal. ArXiv190208736 Eess | 2019 | AMPds2 | I,P,Q,S | No | P | gated dilated CNN (3.25e6) | |
| Bejarano G, DeFazio D, Ramesh A (2019) Deep Latent Generative Models For Energy Disaggregation | 2019 | dataport, REDD | P | No | P | VRNN (not av.) | https://bitbucket.org/gissemari/disaggregation-vrnn/src/master/ |
| Shin C, Joo S, Yim J, et al (2018) Subtask Gated Networks for Non-Intrusive Load Monitoring. ArXiv181106692 Cs Stat | 2018 | REDD, UK-DALE | No | P (on/off) | CNN for subnets (not av.) | ||
| Rafiq H, Zhang H, Li H, Ochani MK (2018) Regularized LSTM Based Deep Learning Model: First Step towards Real-Time Non-Intrusive Load Monitoring. In: 2018 IEEE International Conference on Smart Energy Grid Engineering (SEGE). IEEE, pp 234–239 | 2018 | UK-Dale | P | No | P | LSTM, GRU (not av.) | |
| Martins PBM, Gomes JGRC, Nascimento VB, de Freitas AR (2018) Application of a Deep Learning Generative Model to Load Disaggregation for Industrial Machinery Power Consumption Monitoring. In: 2018 IEEE International Conference on Communications, Control, and Computing Technologies for Smart Grids (SmartGridComm). IEEE, Aalborg, pp 1–6 | 2018 | industrial, see code | P | No | P | gated dilated CNN (155e3) | |
| Murray D, Stankovic L, Stankovic V, et al (2018) Transferability of neural networks approaches for low-rate energy disaggregation. In: 2019 International Conference on Acoustics, Speech, and Signal Processing | 2018 | REDD, REFIT, UK-DALE | P | No | on/off & P, single value | (1)GRU (5e3), (2)CNN (29e6) | |
| Valenti M, Bonfigli R, Principi E, Squartini and S (2018) Exploiting the Reactive Power in Deep Neural Models for Non-Intrusive Load Monitoring. In: 2018 International Joint Conference on Neural Networks (IJCNN). IEEE, Rio de Janeiro, pp 1–8 | 2018 | AmpDS2, UK-DALE | P,Q | Yes | P | CNN dAE () | |
| O. Krystalakos, C. Nalmpantis, and D. Vrakas, C10Sliding Window Approach for Online Energy Disaggregation Using Artificial Neural Networks�, in Proceedings of the 10th Hellenic Conference on Artificial Intelligence, New York, NY, USA, 2018, pp. 7:1�7:6. | 2018 | https://github.com/OdysseasKr/online-nilm | |||||
| Barsim, K.S., Yang, B.: On the Feasibility of Generic Deep Disaggregation for Single-Load Extraction (2018). 1802.02139 | 2018 | UK-DALe | P | No | on/off | CNN dAE (41e6) | |
| Bonfigli R, Felicetti A, Principi E, Fagiani M, Squartini S, Piazza F (2018) Denoising autoencoders for non-intrusive load monitoring: improvements and comparative evaluation. Energy Buildings 158:1461�1474 | 2018 | AMPds, REDD, UK-DALE | Yes | P | CNN dAE () | ||
| Felan Carlo C. Garcia, Christine May C. Creayla, and Erees Queen B. Macabebe, “Development of an Intelligent System for Smart Home Energy Disaggregation Using Stacked Denoising Autoencoders,” Procedia Computer Science, vol. 105, pp. 248–255, 2017. | 2017 | proprietary | P | Yes | P | dAE | |
| J. Kim, T.-T.-H. Le, and H. Kim, Nonintrusive Load Monitoring Based on Advanced Deep Learning and Novel Signature, Computational Intelligence and Neuroscience, 2017. | 2017 | UK-DALE, REDD | Prv. | No | on/off | LSTM | |
| Morgan ES (2017) Applications of deep learning for load disaggregation in residential environments. Master Thesis, Bachelor’s thesis, Universidade Federal do Rio de Janeiro, Rio de Janeiro | 2017 | proprietary, only fridge | P, Q, S | No | P, Q | best: Conv dAE with skip connections (463e3) | |
| C. Zhang, M. Zhong, Z. Wang, N. Goddard, and C. Sutton, �Sequence-to-point learning with neural networks for nonintrusive load monitoring�. arXiv, Dec. 2016. | 2016 | UK-DALE, REDD | P | No | P | CNN (not av.) | |
| W. He and Y. Chai, An Empirical Study on Energy Disaggregation via Deep Learning�. The 2016 2nd International Conference on Artificial Intelligence and Industrial Engineering (AIIE2016), Beijing, China Nov. 2016. | 2016 | UK-DALE | No | P | En-/Decoder, LSTM | ||
| L. Mauch, B. Yang, A novel DNN-HMM-based approach for extracting single loads from aggregate power signals�. In proceedings of the IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP):2384�2388, Mar. 2016 | 2016 | REDD | P | Yes | P | FeedForward and HMM (2e6) | |
| LP. Nascimento, Applications of deep learning techniques on NILM. Phd Thesis, Universidade Federal do Rio de Janeiro (2016) | 2016 | REDD | En-/Decoder à CNN, RNN, Res-Net based | ||||
| Mauch and B. Yang, �A new approach for supervised power disaggregation by using a deep recurrent LSTM network�. In proceedings of the 3rd IEEE Global Conference on Signal and Information Processing (GlobalSIP):63�67, Dec. 2015. | 2015 | REDD | Yes | LSTM | |||
| J. Kelly and W. J. Knottenbelt, Neural NILM: Deep Neural Networks Applied to Energy Disaggregation .CoRR abs/1507.06594 , Aug. 2015. | 2015 | UK-DALE | P | Yes | P | (1)LSTM (1e6), (2)dAE (1-150e6), (3)CNN Rect (28-120e6) | https://github.com/JackKelly/neuralnilm |
| Huss A (2015) Hybrid Model Approach to Appliance Load Disaggregation: Expressive appliance modelling by combining convolutional neural networks and hidden semi Markov models. | 2015 | CNN | . |
Check out "Awsome NILM" to find further NILM resources.