# Non-Intrusive Load Decomposition Based on Instance-Batch Normalization Networks

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Non-Intrusive Load Decomposition Model

#### 2.1. Non-Intrusive Load Decomposition Problem Modeling

#### 2.2. Seq2point Framework

#### 2.3. Flow Chart of Load Decomposition in This Paper

## 3. Seq2point Model Based on IBN-Net Codec Mechanism

#### 3.1. IBN-Net Sub-Module

#### 3.2. Attention Mechanism

## 4. Data Pre-Processing and Experimental Setup

#### 4.1. Dataset Selection

#### 4.2. Electrical Selection

#### 4.3. Data Pre-Processing

#### 4.4. Experimental Setup

#### 4.5. Performance Metrics

## 5. Discussion

#### 5.1. Comparison and Analysis of Experimental Results of the Same House

#### 5.2. Comparison and Analysis of Experimental Results of Different Houses

#### 5.3. Comparison and Analysis of Experimental Results of Transfer Learning

#### 5.4. Ablation Experiment

## 6. Conclusions

## Author Contributions

## Funding

## Data Availability Statement

## Conflicts of Interest

## References

- Zhu, H.; Cao, N.; Deer, H.; Zhang, Z.; Ke, W. Non-intrusive load identification method based on feature-weighted KNN. Electron. Meas. Technol.
**2022**, 45, 70–75. [Google Scholar] - Deng, X.P.; Zhang, G.Q.; Wei, Q.L.; Wei, P.; Li, C.-D. A survey on the non-intrusive load monitoring. Acta Autom. Sin.
**2022**, 48, 644–663. [Google Scholar] - Cui, H.Y.; Cai, J.; Chen, L.; Jiang, C.; Jiang, Y.; Zhang, X. Non-intrusive load fine-grained identification based on color encoding. Power Syst. Technol.
**2022**, 46, 1557–1565. [Google Scholar] - Chen, J.; Wang, X.; Zhang, H. Non-intrusive load recognition using color encoding in edge computing. Chin. J. Sci. Instrum.
**2020**, 41, 12–19. [Google Scholar] - Hart, G.W. Nonintrusive appliance load monitoring. Proc. IEEE
**1992**, 80, 1870–1891. [Google Scholar] [CrossRef] - Kolter, J.Z.; Johnson, M.J. REDD: A public data set for energy disaggregation research. In Proceedings of the Workshop on Data Mining Applications in Sustainability (SIGKDD), San Diego, CA, USA, 21 August 2011; Volume 25, pp. 59–62. [Google Scholar]
- Kelly, J.; Knottenbelt, W. The UK-DALE dataset, domestic appliance-level electricity demand and whole-house demand from five UK homes. Sci. Data
**2015**, 2, 150007. [Google Scholar] [CrossRef] [Green Version] - Batra, N.; Kelly, J.; Parson, O.; Dutta, H.; Knottenbelt, W.; Rogers, A.; Singh, A.; Srivastava, M. NILMTK: An Open Source Toolkit for Non-intrusive Load Monitoring. In Proceedings of the 5th International Conference on Future Energy Systems (ACM e-Energy), Cambridge, UK, 11–13 June 2014; pp. 265–276. [Google Scholar]
- Kelly, J.; Knottenbelt, W. Neural NILM: Deep Neural Networks Applied to Energy Disaggregation. In Proceedings of the 2nd ACM International Conference on Embedded Systems for Energy-Efficient Built Environments, Seoul, Republic of Korea, 4–5 November 2015; pp. 55–64. [Google Scholar]
- Batra, N.; Kukunuri, R.; Pandey, A.; Malakar, R.; Kumar, R.; Krystalakos, O.; Zhong, M.; Meira, P.; Parson, O. A demonstration of reproducible state-of-the-art energy disaggregation using NILMTK. In Proceedings of the 6th ACM International Conference, New York, NY, USA, 13–14 November 2019; pp. 358–359. [Google Scholar]
- Xu, X.; Zhao, S.; Cui, K. Non-intrusive load disaggregate algorithm based on convolutional block attention module. Power Syst. Technol.
**2021**, 45, 3700–3705. [Google Scholar] - Liu, R. Non-Intrusive Load Decomposition Based on Data Augmentation and Deep Learning; Zhejiang University: Hangzhou, China, 2021. [Google Scholar]
- Wu, Z.; Wang, C.; Peng, W.; Liu, W.; Zhang, H. Non-intrusive load monitoring using factorial hidden markov model based on adaptive density peak clustering. Energy Build.
**2021**, 244, 111025. [Google Scholar] [CrossRef] - Yu, C.; Qin, Z.J.; Yang, Y.D. Non- intrusive Load Disaggregation by Improved Factorial Hidden Markov Model Considering ON-OFF Status Recognition. Power Syst. Technol.
**2021**, 45, 4540–4550. [Google Scholar] - Gong, F.; Han, N.; Zhou, Y.; Chen, S.; Li, D.; Tian, S. A SVM Optimized by Particle Swarm Optimization Approach to Load Disaggregation in Non-Intrusive Load Monitoring in Smart Homes. In Proceedings of the 2019 IEEE 3rd Conference on Energy Internet and Energy System Integration (EI2), Changsha, China, 8–10 November 2019; pp. 1793–1797. [Google Scholar]
- Hidiyanto, F.; Halim, A. KNN Methods with Varied K, Distance and training data to disaggregate NILM with similar load characteristic. In Proceedings of the 3rd Asia Pacific Conference on Research in Industrial and Systems Engineering, Depok, Indonesia, 16–17 June 2020; pp. 93–99. [Google Scholar]
- Li, K.; Feng, J.; Xing, Y.; Wang, B. A Self-training Multi-task Attention Method for NILM. In Proceedings of the 2022 IEEE 11th Data Driven Control and Learning Systems Conference, Chengdu, China, 3–5 August 2022; pp. 11–15. [Google Scholar]
- Zhang, C.; Zhong, M.; Wang, Z.; Goddard, N.; Sutton, C. Sequence-to-Point Learning with Neural Networks for Non-Intrusive Load Monitoring. In Proceedings of the National Conference on Artificial Intelligence, New Orleans, LA, USA, 2–7 February 2018; AAAI Press: Palo Alto, CA, USA, 2018; pp. 2604–2611. [Google Scholar]
- D’Incecco, M.; Squartini, S.; Zhong, M. Transfer Learning for Non-Intrusive Load Monitoring. IEEE Trans-Actions Smart Grid
**2020**, 11, 1419–1429. [Google Scholar] [CrossRef] [Green Version] - Piccialli, V.; Sudoso, A.M. Improving Non-Intrusive Load Disaggregation through an Attention-Based Deep Neural Network. Energies
**2021**, 14, 847. [Google Scholar] [CrossRef] - Cao, H.; Shen, X.; Liu, C. Improved infrared target detection algorithm of YOLOv3. J. Electron. Meas. Instrum.
**2020**, 32, 188–194. [Google Scholar] - Liu, J.W.; Zhao, H.D.; Luo, X.L. Research progress on deep learning batch normalization and its related algorithms. Acta Autom. Sin.
**2020**, 46, 1090–1119. [Google Scholar] - Pan, X.; Luo, P.; Shi, J.; Tang, X. Two at Once: Enhancing Learning and Generalization Capacities via IBN-Net. In Proceedings of the Computer Vision—ECCV 2018, Munich, Germany, 8–14 September 2018; pp. 484–500. [Google Scholar]
- Virtsionis-Gkalinikis, N.; Nalmpantis, C.; Vrakas, D. SAED: Self-attentive energy disaggregation. Mach. Learn.
**2021**, 1–20. [Google Scholar] [CrossRef] - Yue, J.R.; Song, Y.Q.; Yang, D.X.; Li, L. Non-Intrusive Load Decomposition Algorithm Based on seq2seq Model. Electrical Measurement and Instrumentation. 2021. Available online: https://kns.cnki.net/kcms/detail/23.1202.TH.20210705.1100.005.html (accessed on 23 February 2023).

**Figure 6.**Example graph of experimental results of the same house (

**a**) Refrigerator (

**b**) Dishwasher (

**c**) Microwave (

**d**) Washing machine.

Metric | Models | RF | WM | DW | MW | Average |
---|---|---|---|---|---|---|

MAE (W) | DAE | 24.071 | 23.352 | 13.631 | 13.376 | 18.608 |

Seq2seq | 23.441 | 15.315 | 8.794 | 9.725 | 14.319 | |

Seq2point | 20.169 | 18.967 | 8.759 | 11.659 | 14.889 | |

LDwA | 19.037 | 13.416 | 8.954 | 8.694 | 12.525 | |

IBN-Net | 15.855 | 9.707 | 7.057 | 8.192 | 10.203 | |

RMSE (W) | DAE | 37.578 | 141.003 | 119.317 | 86.412 | 96.078 |

Seq2seq | 34.437 | 117.376 | 95.186 | 73.740 | 80.185 | |

Seq2point | 34.665 | 118.916 | 88.592 | 83.978 | 90.788 | |

LDwA | 29.154 | 85.112 | 66.712 | 76.037 | 64.254 | |

IBN-Net | 28.529 | 75.163 | 59.491 | 71.365 | 58.637 | |

F1 | DAE | 0.783 | 0.253 | 0.332 | 0.313 | 0.420 |

Seq2seq | 0.792 | 0.533 | 0.608 | 0.454 | 0.597 | |

Seq2point | 0.832 | 0.664 | 0.646 | 0.480 | 0.656 | |

LDwA | 0.905 | 0.822 | 0.672 | 0.689 | 0.772 | |

IBN-Net | 0.916 | 0.871 | 0.753 | 0.819 | 0.840 |

Metric | Model Solutions | IN and BN | Skip Connection | Attention | RF | WM | DW | MW | Average |
---|---|---|---|---|---|---|---|---|---|

MAE (W) | A | √ | 21.243 | 13.445 | 9.386 | 10.538 | 13.653 | ||

B | √ | √ | 24.005 | 15.865 | 11.175 | 12.263 | 15.827 | ||

C | √ | √ | 17.758 | 10.874 | 7.907 | 9.170 | 11.427 | ||

Complete model | √ | √ | √ | 15.855 | 9.707 | 7.057 | 8.192 | 10.203 | |

F1 | A | √ | 0.817 | 0.744 | 0.643 | 0.732 | 0.734 | ||

B | √ | √ | 0.861 | 0.818 | 0.707 | 0.770 | 0.789 | ||

C | √ | √ | 0.887 | 0.845 | 0.732 | 0.798 | 0.815 | ||

Complete model | √ | √ | √ | 0.916 | 0.871 | 0.753 | 0.819 | 0.840 |

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |

© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Wang, M.; Liu, D.; Li, C.
Non-Intrusive Load Decomposition Based on Instance-Batch Normalization Networks. *Energies* **2023**, *16*, 2940.
https://doi.org/10.3390/en16072940

**AMA Style**

Wang M, Liu D, Li C.
Non-Intrusive Load Decomposition Based on Instance-Batch Normalization Networks. *Energies*. 2023; 16(7):2940.
https://doi.org/10.3390/en16072940

**Chicago/Turabian Style**

Wang, Mao, Dandan Liu, and Changzhi Li.
2023. "Non-Intrusive Load Decomposition Based on Instance-Batch Normalization Networks" *Energies* 16, no. 7: 2940.
https://doi.org/10.3390/en16072940