Article
Energy & Fuels
Tamara Todic, Vladimir Stankovic, Lina Stankovic
Summary: With the widespread deployment of smart meters, Non-Intrusive Load Monitoring (NILM) has emerged as a promising application for informing energy management within buildings. However, existing deep learning NILM models have limitations in terms of flexibility and scalability. In this study, an active learning framework is proposed to improve transferability and reduce the cost of labeling, achieving optimal accuracy-labelling effort trade-off. The results demonstrate the potential of this approach in improving the performance of NILM models and reducing computational resources needed.
Article
Energy & Fuels
Xiaochao Guo, Chao Wang, Tao Wu, Ruiheng Li, Houyi Zhu, Huaiqing Zhang
Summary: Results from Non-Intrusive Load Monitoring can be used for energy decomposition and load identification, facilitating effective energy consumption management. However, existing studies often focus on fixed settings, which limits the practical application in real-world scenarios. In this study, we propose a novel appliance detection method that analyzes pattern variations of aggregated power sequences and can determine new appliances without prior information, establishing a basis for load monitoring in scenarios with dynamic changes in load topology. Experimental results show significant improvement in F1 metric compared to event-based approaches.
Article
Energy & Fuels
Yu Liu, Tiancheng E. Song, Xiaolong Sun, Shan Gao, Xueliang Huang
Summary: This paper explores the customized temporal behaviors for load disaggregation, proposing a two-stage framework that integrates probabilistic temporal weights for optimal disaggregation decision. Through comprehensive verifications on a low voltage networks simulator, it is demonstrated that the proposed approach is effective in temporal load feature modeling, achieving higher accuracy and flexibility for the non-intrusive load monitoring problem.
Article
Engineering, Electrical & Electronic
Sagar Verma, Shikha Singh, Angshul Majumdar
Summary: This work introduces a multi-label classification based paradigm for NILM and demonstrates how to consider the temporal variability of input signals in a classification framework. Results on benchmark datasets such as REDD and Pecan Street show considerable improvement over several state-of-the-art shallow and deep techniques.
ELECTRIC POWER SYSTEMS RESEARCH
(2021)
Article
Chemistry, Analytical
Stavros Sykiotis, Maria Kaselimi, Anastasios Doulamis, Nikolaos Doulamis
Summary: This paper proposes an NILM architecture called ELECTRIcity, which utilizes transformer layers and attention mechanisms to accurately estimate the power signal of domestic appliances. Compared to traditional transformer-based architectures, ELECTRIcity addresses the issues of data preprocessing and training time efficiently and outperforms several state-of-the-art methods in predictive accuracy.
Article
Energy & Fuels
Karoline Brucke, Stefan Arens, Jan-Simon Telle, Thomas Steens, Benedikt Hanke, Karsten von Maydell, Carsten Agert
Summary: A new algorithm was developed to extract device profiles from three-phases reactive and active aggregate power measurements in an unsupervised manner, and applied to disaggregate the data and make short-term power predictions with high accuracy.
Article
Chemistry, Analytical
Nasrin Kianpoor, Bjarte Hoff, Trond Ostrem
Summary: This study proposes an adaptive ensemble filtering framework integrated with long short-term memory (LSTM) for identifying flexible loads. The framework, called AEFLSTM, combines filtering techniques and LSTM to filter noise and learn the long-term dependencies of flexible loads. It also searches ensemble filtering techniques to find the best method for disaggregating different loads.
Review
Engineering, Electrical & Electronic
Pascal A. Schirmer, Iosif Mporas
Summary: The rapid development of technology in the electrical energy sector has led to increased electric power needs. This has resulted in the adoption of smart-meters and smart-grids, as well as the rise of Load Monitoring (LM) using Non-Intrusive Load Monitoring (NILM) for appliance-specific energy monitoring. The article provides a review of NILM approaches, groups previously published results, and includes a software implementation of the described NILM approaches.
IEEE TRANSACTIONS ON SMART GRID
(2023)
Article
Computer Science, Information Systems
B. Gowrienanthan, N. Kiruthihan, K. D. I. S. Rathnayake, S. Kiruthikan, V. Logeeshan, S. Kumarawadu, C. Wanigasekara
Summary: Non-Intrusive Load Monitoring (NILM) is a method to determine individual appliance power consumption from overall power consumption. Implementation of NILM for buildings with a three-phase supply has been challenging, but a deep learning-based approach has shown significant improvement in load disaggregation.
Article
Energy & Fuels
Dandan Li, Jiangfeng Li, Xin Zeng, Vladimir Stankovic, Lina Stankovic, Changjiang Xiao, Qingjiang Shi
Summary: This study explores the importance of reducing carbon emissions in buildings to overall greenhouse gas emissions reductions and proposes a method for estimating appliance power consumption based on transfer learning. By utilizing a one-to-many model and incorporating appliance transfer learning and cross-domain transfer learning, this approach effectively estimates the power consumption of all appliances.
Review
Chemistry, Analytical
Maria Kaselimi, Eftychios Protopapadakis, Athanasios Voulodimos, Nikolaos Doulamis, Anastasios Doulamis
Summary: This paper provides a comprehensive review of the development of non-intrusive load monitoring (NILM), focusing on the transition from mathematical modeling to practical application. The paper reviews and analyzes recent research on NILM methods for residential appliances, highlighting the key findings and discussing the research dilemmas in applying NILM methods. The paper emphasizes the need to transform traditional disaggregation models into a practical and trustworthy framework.
Article
Engineering, Electrical & Electronic
Yinyan Liu, Jing Qiu, Jin Ma
Summary: This paper proposes a Scale- and Attention-experts based Multi-task neural network for non-intrusive load monitoring. By utilizing the correlation between tasks and using self-attention mechanism for weighted fusion, the effectiveness and superiority of the model are demonstrated through extensive experimental results.
IEEE TRANSACTIONS ON SMART GRID
(2022)
Article
Engineering, Electrical & Electronic
Mozaffar Etezadifar, Houshang Karimi, Jean Mahseredjian
Summary: Non-intrusive load monitoring (NILM) is an important tool in demand-side management. This study comprehensively investigates the performances of eight clustering algorithms in NILM and analyzes the impact of choosing different input signals on transient states clustering.
ELECTRIC POWER SYSTEMS RESEARCH
(2023)
Article
Energy & Fuels
Nikolaos Virtsionis Gkalinikis, Christoforos Nalmpantis, Dimitris Vrakas
Summary: This research introduces Torch-NILM, an open-source framework aimed at aiding researchers and engineers in utilizing Pytorch for training deep neural networks in the task of energy disaggregation. By standardizing the experimental setup, the framework addresses issues of comparability and reproducibility often encountered in NILM research.
Article
Engineering, Electrical & Electronic
Chao Wang, Zhao Wu, Wenxiong Peng, Weihua Liu, Linyun Xiong, Tao Wu, Lili Yu, Huaiqing Zhang
Summary: NILM is an energy-saving technology that estimates energy consumption information from aggregated power data. This paper presents an Adaptive Factorial Hidden Markov Model (Adaptive-FHMM) to characterize appliance working states transition and predict power consumption. Experimental results demonstrate that the proposed model outperforms state-of-the-art models in various metrics.
INTERNATIONAL JOURNAL OF ELECTRICAL POWER & ENERGY SYSTEMS
(2022)
Article
Computer Science, Information Systems
Megha Gaur, Stephen Makonin, Ivan Bajic, Angshul Majumdar
Proceedings Paper
Engineering, Electrical & Electronic
Megha Gupta, Angshul Majumdar
2016 24TH EUROPEAN SIGNAL PROCESSING CONFERENCE (EUSIPCO)
(2016)